Article(id=1156949464771547889, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1156949362480861758, articleNumber=null, orderNo=null, doi=10.12404/j.issn.1671-1815.2309705, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=1701964800000, receivedDateStr=2023-12-08, revisedDate=1731600000000, revisedDateStr=2024-11-15, acceptedDate=null, acceptedDateStr=null, onlineDate=1753767847483, onlineDateStr=2025-07-29, pubDate=1738944000000, pubDateStr=2025-02-08, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1753767847483, onlineIssueDateStr=2025-07-29, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1753767847483, creator=13701087609, updateTime=1753767847483, updator=13701087609, issue=Issue{id=1156949362480861758, tenantId=1146029695717560320, journalId=1146123166801305609, year='2025', volume='25', issue='4', pageStart='1312', pageEnd='1751', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=0, createTime=1753767823094, creator=13701087609, updateTime=1755171161273, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1162835389472424814, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1156949362480861758, language=EN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1162835389472424815, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1156949362480861758, language=CN, specialIssueTitle=, coverIllustrator=, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=1580, endPage=1586, ext={EN=ArticleExt(id=1156949465396499201, articleId=1156949464771547889, tenantId=1146029695717560320, journalId=1146123166801305609, language=EN, title=Fire Detection Algorithm Combining Adaptive Gaussian Mixture Model and ResDN, columnId=1156262729162810294, journalTitle=Science Technology and Engineering, columnName=Papers·Automation and Computational Technology, runingTitle=null, highlight=null, articleAbstract=

A lightweight and efficient two-stage video flame detection algorithm was designed to address issues of high false positive rates, poor adaptability, and low efficiency in complex scenes. In the first stage, an improved adaptive Gaussian mixture model (AGMM) was employed for rapid background modeling of video image sequences. Suspicious candidate regions were extracted from the sequences by leveraging the flickering and surging characteristics of flames. In the second stage, a residual deep normalization and convolutional neural network (ResDN) was used to discriminate these suspicious candidate regions. A simplified residual block was introduced to replace the original convolutional layers for a lightweight design, enabling accurate flame detection and localization. Compared with traditional classification algorithms, the proposed two-stage video flame detection algorithm effectively overcomes environmental interference in complex scenes, rapidly and accurately identifies flames, and demonstrates higher detection rates and adaptability.

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Wen-biao WANG, Qi-heng SHI, You-wei HAO), CN=ArticleExt(id=1156949505443714044, articleId=1156949464771547889, tenantId=1146029695717560320, journalId=1146123166801305609, language=CN, title=基于自适应高斯混合模型与ResDN的火焰检测算法, columnId=1156262729783567290, journalTitle=科学技术与工程, columnName=论文·自动化技术、计算机技术, runingTitle=null, highlight=null, articleAbstract=

针对火焰检测算法在复杂场景下误检率高、算法适应性差、效率低等问题,设计一种轻量高效的两阶段视频火焰检测算法。第一阶段采用改进的自适应高斯混合模型(adaptive gaussian mixture model,AGMM)对视频图像序列进行快速背景建模,利用火焰的闪烁和涌动特性,提取出序列中的可疑候选区域。第二阶段使用残差深度归一化卷积神经网络 (residual deep normalization and convolutional neural network, ResDN)对可疑候选区域进行判别,并引入简化的残差块替换原有的卷积层进行轻量化设计,实现对火焰的检测与定位。相比于传统分类算法,所设计的两阶段视频火焰检测算法能够有效克服复杂场景下的环境干扰,准确快速地识别火焰,具有更高的检测率和适应性。

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=6hDO3ce9mAK6ZHpEdcH9Yg==, magXml=+3P5cV4sAW+lZQLv6G0q+Q==, pdfUrl=null, pdf=jdgch6n2r33FKb+GaOINeg==, pdfFileSize=7385779, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=bJLQw4syjCxcsJjIpTGVFw==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=Vn5INhd2LS25GhDp/YaKiQ==, mapNumber=null, authorCompany=null, fund=null, authors=

王文标(1982—),男,汉族,天津人,博士,讲师。研究方向:工业过程控制、机器视觉、图像分析与处理。E-mail:

, authorsList=王文标, 时启衡, 郝友维)}, authors=[Author(id=1225944413139026781, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=dmuwwb@dlmu.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1225944413277438827, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, authorId=1225944413139026781, language=EN, stringName=Wen-biao WANG, firstName=Wen-biao, middleName=null, lastName=WANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1225944413386490740, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, authorId=1225944413139026781, language=CN, stringName=王文标, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=大连海事大学船舶电气工程学院, 大连 116026, bio={"content":"

王文标(1982—),男,汉族,天津人,博士,讲师。研究方向:工业过程控制、机器视觉、图像分析与处理。E-mail:

"}, bioImg=null, bioContent=

王文标(1982—),男,汉族,天津人,博士,讲师。研究方向:工业过程控制、机器视觉、图像分析与处理。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1225944413000614739, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, xref=null, ext=[AuthorCompanyExt(id=1225944413017391957, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China), AuthorCompanyExt(id=1225944413034169175, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=大连海事大学船舶电气工程学院, 大连 116026)])]), Author(id=1225944413482959741, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1225944413583623044, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, authorId=1225944413482959741, language=EN, stringName=Qi-heng SHI, firstName=Qi-heng, middleName=null, lastName=SHI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1225944413680092047, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, authorId=1225944413482959741, language=CN, stringName=时启衡, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=大连海事大学船舶电气工程学院, 大连 116026, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1225944413000614739, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, xref=null, ext=[AuthorCompanyExt(id=1225944413017391957, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China), AuthorCompanyExt(id=1225944413034169175, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=大连海事大学船舶电气工程学院, 大连 116026)])]), Author(id=1225944413814309796, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1225944414036607921, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, authorId=1225944413814309796, language=EN, stringName=You-wei HAO, firstName=You-wei, middleName=null, lastName=HAO, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1225944414183408579, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, authorId=1225944413814309796, language=CN, stringName=郝友维, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=大连海事大学船舶电气工程学院, 大连 116026, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1225944413000614739, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, xref=null, ext=[AuthorCompanyExt(id=1225944413017391957, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China), AuthorCompanyExt(id=1225944413034169175, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=大连海事大学船舶电气工程学院, 大连 116026)])])], keywords=[Keyword(id=1225944414401512408, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, orderNo=1, keyword=flame detection), Keyword(id=1225944414535730155, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, orderNo=2, keyword=adaptive Gaussian mixture model(AGMM)), Keyword(id=1225944416070844418, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, orderNo=3, keyword=residual deep normalization and convolutional neural network(ResDN)), Keyword(id=1225944416368640016, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, orderNo=4, keyword=machine vision), Keyword(id=1225944416595132449, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, orderNo=5, keyword=deep learning), Keyword(id=1225944416725155881, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, orderNo=1, keyword=火焰检测), Keyword(id=1225944416834207796, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, orderNo=2, keyword=自适应高斯混合模型(AGMM)), Keyword(id=1225944416981008450, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, orderNo=3, keyword=残差深度归一化卷积神经网络 (ResDN)), Keyword(id=1225944417148780625, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, orderNo=4, keyword=机器视觉), Keyword(id=1225944417308164191, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, orderNo=5, keyword=深度学习)], refs=[Reference(id=1225944422311969255, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2019, volume=15, issue=5, pageStart=185, pageEnd=191, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=陈志芬, 黄靖玲, 李亚, journalName=中国安全生产科学技术, refType=null, unstructuredReference=陈志芬, 黄靖玲, 李亚. 适应城市消防规划需求的火灾风险评估研究[J]. 中国安全生产科学技术, 2019, 15(5): 185-191., articleTitle=适应城市消防规划需求的火灾风险评估研究, refAbstract=null), Reference(id=1225944422475547129, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2019, volume=15, issue=5, pageStart=185, pageEnd=191, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=Chen Zhifen, Huang Jingling, Li Ya, journalName=Journal of Safety Science and Technology, refType=null, unstructuredReference=Chen Zhifen, Huang Jingling, Li Ya. Study on fire risk assessment adapting to requirements of urban fire protection planning[J]. Journal of Safety Science and Technology, 2019, 15(5): 185-191., articleTitle=Study on fire risk assessment adapting to requirements of urban fire protection planning, refAbstract=null), Reference(id=1225944422647513606, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=42, issue=3, pageStart=416, pageEnd=421, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=唐声强, 黄晓家, 谢水波, journalName=消防科学与技术, refType=null, unstructuredReference=唐声强, 黄晓家, 谢水波, 等. 火灾对我国家庭财产保险需求影响的实证研究[J]. 消防科学与技术, 2023, 42(3): 416-421., articleTitle=火灾对我国家庭财产保险需求影响的实证研究, refAbstract=null), Reference(id=1225944422848840223, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=42, issue=3, pageStart=416, pageEnd=421, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=Tang Shengqiang, Huang Xiaojia, Xie Shuibo, journalName=Fire Science and Technology, refType=null, unstructuredReference=Tang Shengqiang, Huang Xiaojia, Xie Shuibo, et al. An empirical study on the impact of fire on the demand of family property insurance in China[J]. Fire Science and Technology, 2023, 42(3): 416-421., articleTitle=An empirical study on the impact of fire on the demand of family property insurance in China, refAbstract=null), Reference(id=1225944423016612399, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=42, issue=1, pageStart=53, pageEnd=57, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=侯逸臣, 何建, 王琳, journalName=计算技术与自动化, refType=null, unstructuredReference=侯逸臣, 何建, 王琳, 等. 基于多分类器融合的林火检测方法研究[J]. 计算技术与自动化, 2023, 42(1): 53-57., articleTitle=基于多分类器融合的林火检测方法研究, refAbstract=null), Reference(id=1225944423222133321, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=42, issue=1, pageStart=53, pageEnd=57, url=null, language=null, rfNumber=[3], rfOrder=5, authorNames=Hou Yichen, He Jian, Wang Lin, journalName=Computing Technology and Automation, refType=null, unstructuredReference=Hou Yichen, He Jian, Wang Lin, et al. Research on forest fire detection method based on multi-classifier fusion[J]. Computing Technology and Automation, 2023, 42(1): 53-57., articleTitle=Research on forest fire detection method based on multi-classifier fusion, refAbstract=null), Reference(id=1225944423624786534, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=18, issue=1, pageStart=1, pageEnd=6, url=null, language=null, rfNumber=[4], rfOrder=6, authorNames=卞永明, 高飞, 李梦如, journalName=中国工程机械学报, refType=null, unstructuredReference=卞永明, 高飞, 李梦如, 等. 结合K-means++聚类和颜色几何特征的火焰检测方法[J]. 中国工程机械学报, 2020, 18(1): 1-6., articleTitle=结合K-means++聚类和颜色几何特征的火焰检测方法, refAbstract=null), Reference(id=1225944424962769529, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=18, issue=1, pageStart=1, pageEnd=6, url=null, language=null, rfNumber=[4], rfOrder=7, authorNames=Bian Yongming, Gao Fei, Li Mengru, journalName=Chinese Journal of Construction Machinery, refType=null, unstructuredReference=Bian Yongming, Gao Fei, Li Mengru, et al. Fire detection method using K-means++ clustering and features of mixed color and geometry[J]. Chinese Journal of Construction Machinery, 2020, 18(1): 1-6., articleTitle=Fire detection method using K-means++ clustering and features of mixed color and geometry, refAbstract=null), Reference(id=1225944425088598663, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=39, issue=2, pageStart=247, pageEnd=249, url=null, language=null, rfNumber=[5], rfOrder=8, authorNames=葛珺, 王丽, 李伟, journalName=消防科学与技术, refType=null, unstructuredReference=葛珺, 王丽, 李伟, 等. 基于可见光图像的火焰识别与提取[J]. 消防科学与技术, 2020, 39(2): 247-249, 266., articleTitle=基于可见光图像的火焰识别与提取, refAbstract=null), Reference(id=1225944425252176543, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=39, issue=2, pageStart=247, pageEnd=251, url=null, language=null, rfNumber=[5], rfOrder=9, authorNames=Ge Jun, Wang Li, Li Wei, journalName=Fire Science and Technology, refType=null, unstructuredReference=Ge Jun, Wang Li, Li Wei, et al. Recognition and extraction of flames based on visible light images[J]. Fire Science and Technology, 2020, 39(2): 247-251, 266., articleTitle=Recognition and extraction of flames based on visible light images, refAbstract=null), Reference(id=1225944425386394288, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=35, issue=1, pageStart=35, pageEnd=52, url=null, language=null, rfNumber=[6], rfOrder=10, authorNames=曹江涛, 秦跃雁, 姬晓飞, journalName=数据采集与处理, refType=null, unstructuredReference=曹江涛, 秦跃雁, 姬晓飞. 基于视频的火焰检测算法综述[J]. 数据采集与处理, 2020, 35(1): 35-52., articleTitle=基于视频的火焰检测算法综述, refAbstract=null), Reference(id=1225944425625469639, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=35, issue=1, pageStart=35, pageEnd=52, url=null, language=null, rfNumber=[6], rfOrder=11, authorNames=Cao Jiangtao, Qin Yueyan, Ji Xiaofei, journalName=Journal of Data Acquisition & Processing, refType=null, unstructuredReference=Cao Jiangtao, Qin Yueyan, Ji Xiaofei. Review on video based flame detection algorithm[J]. Journal of Data Acquisition & Processing, 2020, 35(1): 35-52., articleTitle=Review on video based flame detection algorithm, refAbstract=null), Reference(id=1225944425768075994, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2017, volume=17, issue=25, pageStart=268, pageEnd=273, url=null, language=null, rfNumber=[7], rfOrder=12, authorNames=宋宁, 强彦, 董林佳, journalName=科学技术与工程, refType=null, unstructuredReference=宋宁, 强彦, 董林佳. 林火监测中基于视觉的火焰检测方法[J]. 科学技术与工程, 2017, 17(25): 268-273., articleTitle=林火监测中基于视觉的火焰检测方法, refAbstract=null), Reference(id=1225944425881322224, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2017, volume=17, issue=25, pageStart=268, pageEnd=273, url=null, language=null, rfNumber=[7], rfOrder=13, authorNames=Song Ning, Qiang Yan, Dong Linjia, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Song Ning, Qiang Yan, Dong Linjia. Vision based flame detection method in forest fire monitoring[J]. Science Technology and Engineering, 2017, 17(25): 268-273., articleTitle=Vision based flame detection method in forest fire monitoring, refAbstract=null), Reference(id=1225944426191700734, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2017, volume=37, issue=7, pageStart=1989, pageEnd=1993, url=null, language=null, rfNumber=[8], rfOrder=14, authorNames=席廷宇, 邱选兵, 孙冬远, journalName=计算机应用, refType=null, unstructuredReference=席廷宇, 邱选兵, 孙冬远, 等. 多特征量对数回归的火焰快速识别算法[J]. 计算机应用, 2017, 37(7): 1989-1993., articleTitle=多特征量对数回归的火焰快速识别算法, refAbstract=null), Reference(id=1225944426296558347, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2017, volume=37, issue=7, pageStart=1989, pageEnd=1993, url=null, language=null, rfNumber=[8], rfOrder=15, authorNames=Xi Tingyu, Qiu Xuanbing, Sun Dongyuan, journalName=Journal of Computer Applications, refType=null, unstructuredReference=Xi Tingyu, Qiu Xuanbing, Sun Dongyuan, et al. Fast fire flame recognition algorithm based on multi-feature logarithmic regression[J]. Journal of Computer Applications, 2017, 37(7): 1989-1993., articleTitle=Fast fire flame recognition algorithm based on multi-feature logarithmic regression, refAbstract=null), Reference(id=1225944426388833046, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=7, issue=1, pageStart=263, pageEnd=275, url=null, language=null, rfNumber=[9], rfOrder=16, authorNames=Wang T, Bu L, Yang Z, journalName=IEEE/CAA Journal of Automatica Sinica, refType=null, unstructuredReference=Wang T, Bu L, Yang Z, et al. A new fire detection method using a multi-expert system based on color dispersion, similarity and centroid motion in indoor environment[J]. IEEE/CAA Journal of Automatica Sinica, 2020, 7(1): 263-275., articleTitle=A new fire detection method using a multi-expert system based on color dispersion, similarity and centroid motion in indoor environment, refAbstract=null), Reference(id=1225944426544022303, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=8, issue=null, pageStart=81904, pageEnd=81917, url=null, language=null, rfNumber=[10], rfOrder=17, authorNames=Xie Y K, Zhu J, Cao Y G, journalName=IEEE Access, refType=null, unstructuredReference=Xie Y K, Zhu J, Cao Y G, et al. Efficient video fire detection exploiting motion-flicker-based dynamic features and deep static features[J]. IEEE Access, 2020, 8: 81904-81917., articleTitle=Efficient video fire detection exploiting motion-flicker-based dynamic features and deep static features, refAbstract=null), Reference(id=1225944426703405871, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=20, issue=3, pageStart=1113, pageEnd=1117, url=null, language=null, rfNumber=[11], rfOrder=18, authorNames=吴雪, 宋晓茹, 高嵩, journalName=科学技术与工程, refType=null, unstructuredReference=吴雪, 宋晓茹, 高嵩, 等. 基于数据增强的卷积神经网络火灾识别[J]. 科学技术与工程, 2020, 20(3): 1113-1117., articleTitle=基于数据增强的卷积神经网络火灾识别, refAbstract=null), Reference(id=1225944427022173011, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=20, issue=3, pageStart=1113, pageEnd=1117, url=null, language=null, rfNumber=[11], rfOrder=19, authorNames=Wu Xue, Song Xiaoru, Gao Song, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Wu Xue, Song Xiaoru, Gao Song, et al. Convolution neural network based on data enhancement for fire identification[J]. Science Technology and Engineering, 2020, 20(3): 1113-1117., articleTitle=Convolution neural network based on data enhancement for fire identification, refAbstract=null), Reference(id=1225944427236082530, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2019, volume=19, issue=32, pageStart=236, pageEnd=243, url=null, language=null, rfNumber=[12], rfOrder=20, authorNames=杨剑, 刘方涛, 张涛, journalName=科学技术与工程, refType=null, unstructuredReference=杨剑, 刘方涛, 张涛, 等. 基于改进型残差网络烟雾图像识别[J]. 科学技术与工程, 2019, 19(32): 236-243., articleTitle=基于改进型残差网络烟雾图像识别, refAbstract=null), Reference(id=1225944427391271790, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2019, volume=19, issue=32, pageStart=236, pageEnd=243, url=null, language=null, rfNumber=[12], rfOrder=21, authorNames=Yang Jian, Liu fangtao, Zhang Tao, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Yang Jian, Liu fangtao, Zhang Tao, et al. Research on smoke image recognition based on improved residual network[J]. Science Technology and Engineering, 2019, 19(32): 236-243., articleTitle=Research on smoke image recognition based on improved residual network, refAbstract=null), Reference(id=1225944427512906615, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=40, issue=3, pageStart=67, pageEnd=74, url=null, language=null, rfNumber=[13], rfOrder=22, authorNames=张旭彤, 胡鹏, 赵鑫, journalName=微电子学与计算机, refType=null, unstructuredReference=张旭彤, 胡鹏, 赵鑫, 等. 火灾预警中基于YOLOv5的火源智能检测定位方法[J]. 微电子学与计算机, 2023, 40(3): 67-74., articleTitle=火灾预警中基于YOLOv5的火源智能检测定位方法, refAbstract=null), Reference(id=1225944427655512968, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=40, issue=3, pageStart=67, pageEnd=74, url=null, language=null, rfNumber=[13], rfOrder=23, authorNames=Zhang Xutong, Hu Peng, Zhao Xin, journalName=Microelectronics & Computer, refType=null, unstructuredReference=Zhang Xutong, Hu Peng, Zhao Xin, et al. Intelligent detection and location method of fire source based on YOLOv5 in fire warning[J]. Microelectronics & Computer, 2023, 40(3): 67-74., articleTitle=Intelligent detection and location method of fire source based on YOLOv5 in fire warning, refAbstract=null), Reference(id=1225944427802313624, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=49, issue=8, pageStart=291, pageEnd=301, url=null, language=null, rfNumber=[14], rfOrder=24, authorNames=陈露萌, 曹彦彦, 黄民, journalName=计算机工程, refType=null, unstructuredReference=陈露萌, 曹彦彦, 黄民, 等. 基于改进YOLOv5的火焰检测方法[J]. 计算机工程, 2023, 49(8): 291-301, 309., articleTitle=基于改进YOLOv5的火焰检测方法, refAbstract=null), Reference(id=1225944428028806060, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2023, volume=49, issue=8, pageStart=291, pageEnd=301, url=null, language=null, rfNumber=[14], rfOrder=25, authorNames=Chen Lumeng, Cao Yanyan, Huang Min, journalName=Computer Engineering, refType=null, unstructuredReference=Chen Lumeng, Cao Yanyan, Huang Min, et al. Flame Detection Method Based on Improved YOLOv5[J]. Computer Engineering, 2023, 49(8): 291-301, 309., articleTitle=Flame Detection Method Based on Improved YOLOv5, refAbstract=null), Reference(id=1225944429400343482, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2024, volume=24, issue=8, pageStart=3298, pageEnd=3307, url=null, language=null, rfNumber=[15], rfOrder=26, authorNames=谢康康, 朱文忠, 肖顺兴, journalName=科学技术与工程, refType=null, unstructuredReference=谢康康, 朱文忠, 肖顺兴, 等. 一种改进YOLOX_S的火焰烟雾检测算法[J]. 科学技术与工程, 2024, 24(8): 3298-3307., articleTitle=一种改进YOLOX_S的火焰烟雾检测算法, refAbstract=null), Reference(id=1225944429605864388, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2024, volume=24, issue=8, pageStart=3298, pageEnd=3307, url=null, language=null, rfNumber=[15], rfOrder=27, authorNames=Xie Kangkang, Zhu Wenzhong, Xiao Shunxing, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Xie Kangkang, Zhu Wenzhong, Xiao Shunxing, et al. Flame smoke detection algorithm based on improved YOLOX_S[J]. Science Technology and Engineering, 2024, 24(8): 3298-3307., articleTitle=Flame smoke detection algorithm based on improved YOLOX_S, refAbstract=null), Reference(id=1225944429882688471, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2021, volume=36, issue=6, pageStart=1276, pageEnd=1285, url=null, language=null, rfNumber=[16], rfOrder=28, authorNames=孙维亚, 陈恺鑫, 吴铭, journalName=数据采集与处理, refType=null, unstructuredReference=孙维亚, 陈恺鑫, 吴铭, 等. 融合运动特征的高效视频火焰检测算法[J]. 数据采集与处理, 2021, 36(6): 1276-1285., articleTitle=融合运动特征的高效视频火焰检测算法, refAbstract=null), Reference(id=1225944430025294819, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2021, volume=36, issue=6, pageStart=1276, pageEnd=1285, url=null, language=null, rfNumber=[16], rfOrder=29, authorNames=Sun Weiya, Chen kaixin, Wu Ming, journalName=Journal of Data Acquisition & Processing, refType=null, unstructuredReference=Sun Weiya, Chen kaixin, Wu Ming, et al. An efficient video flame detection algorithm integrating motion features[J]. Journal of Data Acquisition & Processing, 2021, 36(6): 1276-1285., articleTitle=An efficient video flame detection algorithm integrating motion features, refAbstract=null), Reference(id=1225944430276953082, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2019, volume=40, issue=6, pageStart=874, pageEnd=878, url=null, language=null, rfNumber=[17], rfOrder=30, authorNames=吕苗苗, 孙建明, journalName=半导体光电, refType=null, unstructuredReference=吕苗苗, 孙建明. 基于改进高斯混合模型的运动图像目标检测算法[J]. 半导体光电, 2019, 40(6): 874-878, 885., articleTitle=基于改进高斯混合模型的运动图像目标检测算法, refAbstract=null), Reference(id=1225944430436335621, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2019, volume=40, issue=6, pageStart=874, pageEnd=878, url=null, language=null, rfNumber=[17], rfOrder=31, authorNames=Lü Miaomiao, Sun Jianming, journalName=Semiconductor Optoelectronics, refType=null, unstructuredReference= Miaomiao, Sun Jianming. Moving image target detection based on modified Gaussian mixture model[J]. Semiconductor Optoelectronics, 2019, 40(6): 874-878, 885., articleTitle=Moving image target detection based on modified Gaussian mixture model, refAbstract=null), Reference(id=1225944430629273614, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2022, volume=33, issue=4, pageStart=403, pageEnd=413, url=null, language=null, rfNumber=[18], rfOrder=32, authorNames=侯涛, 宝才文, 陈燕楠, journalName=光电子·激光, refType=null, unstructuredReference=侯涛, 宝才文, 陈燕楠. 基于自适应高斯混合模型的铁轨异物入侵检测研究[J]. 光电子·激光, 2022, 33(4): 403-413., articleTitle=基于自适应高斯混合模型的铁轨异物入侵检测研究, refAbstract=null), Reference(id=1225944430738325533, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2022, volume=33, issue=4, pageStart=403, pageEnd=413, url=null, language=null, rfNumber=[18], rfOrder=33, authorNames=Hou Tao, Bao Caiwen, Chen Yannan, journalName=Journal of Optoelectronics·Laser, refType=null, unstructuredReference=Hou Tao, Bao Caiwen, Chen Yannan. Research on detection of foreign object intrusion in railroad tracks based on AGMM[J]. Journal of Optoelectronics·Laser, 2022, 33(4): 403-413., articleTitle=Research on detection of foreign object intrusion in railroad tracks based on AGMM, refAbstract=null), Reference(id=1225944430843183142, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=20, issue=15, pageStart=6141, pageEnd=6150, url=null, language=null, rfNumber=[19], rfOrder=34, authorNames=李笑, 杨宇, 徐一鸣, journalName=科学技术与工程, refType=null, unstructuredReference=李笑, 杨宇, 徐一鸣. 四帧间差分和改进混合高斯模型对运动目标的检测[J]. 科学技术与工程, 2020, 20(15): 6141-6150., articleTitle=四帧间差分和改进混合高斯模型对运动目标的检测, refAbstract=null), Reference(id=1225944430952235057, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2020, volume=20, issue=15, pageStart=6141, pageEnd=6150, url=null, language=null, rfNumber=[19], rfOrder=35, authorNames=Li Xiao, Yang Yu, Xu Yiming, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Li Xiao, Yang Yu, Xu Yiming. Detection of moving targets by four-frame difference and modified gaussian mixture model[J]. Science Technology and Engineering, 2020, 20(15): 6141-6150., articleTitle=Detection of moving targets by four-frame difference and modified gaussian mixture model, refAbstract=null), Reference(id=1225944431099035704, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2017, volume=5, issue=null, pageStart=6833, pageEnd=6841, url=null, language=null, rfNumber=[20], rfOrder=36, authorNames=Yuan F N, Xia X, Shi J T, journalName=IEEE Access, refType=null, unstructuredReference=Yuan F N, Xia X, Shi J T, et al. Non-linear dimensionality reduction and gaussian process based classification method for smoke detection[J]. IEEE Access, 2017, 5: 6833-6841., articleTitle=Non-linear dimensionality reduction and gaussian process based classification method for smoke detection, refAbstract=null), Reference(id=1225944431258419271, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, doi=null, pmid=null, pmcid=null, year=2012, volume=null, issue=null, pageStart=362, pageEnd=367, url=null, language=null, rfNumber=[21], rfOrder=37, authorNames=Deok-Yeon K, Joon-Young K, Chul K B, journalName=IEEE International Conference on Multimedia and Expo, refType=null, unstructuredReference=Deok-Yeon K, Joon-Young K, Chul K B, et al. Human detection using wavelet-based CS-LBP and a cascade of random forests[C]// IEEE International Conference on Multimedia and Expo. Melbourne: IEEE, 2012: 362-367., articleTitle=Human detection using wavelet-based CS-LBP and a cascade of random forests, refAbstract=null)], funds=[Fund(id=1225944421817041333, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, awardId=52071047, language=CN, fundingSource=国家自然科学基金(52071047), fundOrder=null, country=null), Fund(id=1225944421947064771, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, awardId=62073054, language=CN, fundingSource=国家自然科学基金(62073054), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1225944413000614739, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, xref=null, ext=[AuthorCompanyExt(id=1225944413017391957, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China), AuthorCompanyExt(id=1225944413034169175, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, companyId=1225944413000614739, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=大连海事大学船舶电气工程学院, 大连 116026)])], figs=[ArticleFig(id=1225944417748566158, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Fig.1, caption=Adaptive gaussian mixture model target extraction flowchart, figureFileSmall=YUdNhcqWD3RVNelLtxXkMQ==, figureFileBig=HbNM25tIHe7oXh2rIacqOA==, tableContent=null), ArticleFig(id=1225944417928921250, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=图1, caption=自适应高斯混合模型目标提取流程图, figureFileSmall=YUdNhcqWD3RVNelLtxXkMQ==, figureFileBig=HbNM25tIHe7oXh2rIacqOA==, tableContent=null), ArticleFig(id=1225944418260271290, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Fig.2, caption=ResDN network structure diagram, figureFileSmall=3VMlQbi9JgOWepJlBjqhuA==, figureFileBig=sgglnbHcXzq9pt+kEea0kQ==, tableContent=null), ArticleFig(id=1225944418402877637, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=图2, caption=ResDN结构图, figureFileSmall=3VMlQbi9JgOWepJlBjqhuA==, figureFileBig=sgglnbHcXzq9pt+kEea0kQ==, tableContent=null), ArticleFig(id=1225944418511929552, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Fig.3, caption=Simplified residual block structure diagram, figureFileSmall=+uZL/s9RjJixm8qE5dd6rw==, figureFileBig=kvLXWyfvelKcQgxuS0EOEg==, tableContent=null), ArticleFig(id=1225944418637758692, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=图3, caption=简化的残差块结构图, figureFileSmall=+uZL/s9RjJixm8qE5dd6rw==, figureFileBig=kvLXWyfvelKcQgxuS0EOEg==, tableContent=null), ArticleFig(id=1225944418776170742, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Fig.4, caption=Background modeling results presentation, figureFileSmall=/Wkv0/joP4W4XBJdjO00fg==, figureFileBig=V/dcxQOyP/YK2OrRFcRjmA==, tableContent=null), ArticleFig(id=1225944418914582798, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=图4, caption=背景建模结果展示, figureFileSmall=/Wkv0/joP4W4XBJdjO00fg==, figureFileBig=V/dcxQOyP/YK2OrRFcRjmA==, tableContent=null), ArticleFig(id=1225944419023634714, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Fig.5, caption=Training results display, figureFileSmall=qm+sbIpypDzQNx6Vk8OuDA==, figureFileBig=3PXU8Sh3tWerpHbReeh0LA==, tableContent=null), ArticleFig(id=1225944420483252526, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=图5, caption=训练结果展示

Epoch为训练轮数

, figureFileSmall=qm+sbIpypDzQNx6Vk8OuDA==, figureFileBig=3PXU8Sh3tWerpHbReeh0LA==, tableContent=null), ArticleFig(id=1225944420667801918, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Table 1, caption=

ResDN network structure

, figureFileSmall=null, figureFileBig=null, tableContent=
输入层 类型 数量 卷积核尺寸 步长
Input
DN1 DN 32 3×3 1
Res1 Residual_unit 32 3×3 1
DN2 DN 64 3×3 1
MaxPooling1 Pool 3×3 2
Res2 Residual_unit 64 3×3 1
Res3 Residual_unit 64 3×3 1
MaxPooling2 Pool 2×2 2
DN3 DN 384 3×3 1
Res4 Residual_unit 384 3×3 1
DN4 DN 256 3×3 1
MaxPooling3 Pool 2×2 2
Flatten
Dense1 Dense 2 048
Dense2 Dense 2 048
Output Dense 2
), ArticleFig(id=1225944421074649435, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=表1, caption=

ResDN网络结构

, figureFileSmall=null, figureFileBig=null, tableContent=
输入层 类型 数量 卷积核尺寸 步长
Input
DN1 DN 32 3×3 1
Res1 Residual_unit 32 3×3 1
DN2 DN 64 3×3 1
MaxPooling1 Pool 3×3 2
Res2 Residual_unit 64 3×3 1
Res3 Residual_unit 64 3×3 1
MaxPooling2 Pool 2×2 2
DN3 DN 384 3×3 1
Res4 Residual_unit 384 3×3 1
DN4 DN 256 3×3 1
MaxPooling3 Pool 2×2 2
Flatten
Dense1 Dense 2 048
Dense2 Dense 2 048
Output Dense 2
), ArticleFig(id=1225944421166924137, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Table 2, caption=

Sample detection rate

, figureFileSmall=null, figureFileBig=null, tableContent=
视频集 检测率
HSV GMM AGMM(本文模型)
Video1 0.546 0.854 0.926
Video2 0.834 0.608 0.644
Video3 0.816 0.854 0.906
Video4 0.912 0.822 0.906
Video5 0.924 0.938 0.930
), ArticleFig(id=1225944421338890619, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=表2, caption=

样本检测率

, figureFileSmall=null, figureFileBig=null, tableContent=
视频集 检测率
HSV GMM AGMM(本文模型)
Video1 0.546 0.854 0.926
Video2 0.834 0.608 0.644
Video3 0.816 0.854 0.906
Video4 0.912 0.822 0.906
Video5 0.924 0.938 0.930
), ArticleFig(id=1225944421464719756, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=EN, label=Table 3, caption=

ComparativeExperiment

, figureFileSmall=null, figureFileBig=null, tableContent=
算法 准确率 检出率 虚警率 模型尺寸/MB
ResDN 0.972 99 0.969 42 0.022 94 97.8
DNCNN 0.969 04 0.964 48 0.025 75 195.0
Res18 0.970 56 0.970 89 0.030 95 46.9
Res50 0.969 95 0.966 00 0.025 51 170.0
VGG16 0.933 64 0.895 74 0.023 16 526.0
), ArticleFig(id=1225944421582160281, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1156949464771547889, language=CN, label=表3, caption=

对比实验

, figureFileSmall=null, figureFileBig=null, tableContent=
算法 准确率 检出率 虚警率 模型尺寸/MB
ResDN 0.972 99 0.969 42 0.022 94 97.8
DNCNN 0.969 04 0.964 48 0.025 75 195.0
Res18 0.970 56 0.970 89 0.030 95 46.9
Res50 0.969 95 0.966 00 0.025 51 170.0
VGG16 0.933 64 0.895 74 0.023 16 526.0
)], attaches=null, journal=Journal(id=1146119176004939786, delFlag=0, nameCn=科学技术与工程, nameEn=Science Technology and Engineering, nameHistory1=null, nameHistory2=null, issn=1671-1815, eissn=, cn=11-4688/T, coden=null, periodic=4, language=CN, oaType=是, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=UKU/O7GSka5polgCTkbIIw==, journalPrice=null, startedYear=null, abbrevIsoEn=Sci Technol Eng, journalRemark=null, publicationField=null, createdTime=null, updatedTime=1754445529766, createdBy=null, updatedBy=13701087609, firstLetterCn=S, firstLetterEn=S, subjectCode=Natural Sciences, subjectName=自然科学, subjectCodeEn=Natural Sciences, subjectNameEn=null, picCn=UKU/O7GSka5polgCTkbIIw==, picEn=5hwlULoNwcbj3xUmVi9MAQ==, jcr=null, cjcr=null, exts=[JournalExt(id=1159791870395564357, language=CN, name=科学技术与工程, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.stae.com.cn/jsygc/home, createdTime=1754445529793, updatedTime=1754445529793, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=http://www.stae.com.cn/jsygc/site/menus/20090429150146001, submissionAuthorUrl=http://www.stae.com.cn/jsygc/author/login, submissionEditorUrl=http://www.stae.com.cn/jsygc/editor/login, submissionReviewUrl=http://www.stae.com.cn/jsygc/reviewer/login, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1159791870441701702, language=EN, name=Science Technology and Engineering, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.stae.com.cn/jsygc/home, createdTime=1754445529804, updatedTime=1754445529804, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.stae.com.cn/jsygc/author/login, submissionEditorUrl=http://www.stae.com.cn/jsygc/editor/login, submissionReviewUrl=http://www.stae.com.cn/jsygc/reviewer/login, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1146123166801305609, websiteList=[Website(id=1148243202391400884, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123166801305609, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/kxjsygc/CN, language=CN, createTime=1751692112777, createBy=18614031015, updateTime=1753520965431, updateBy=18614031015, name=科学技术与工程-中文站点, tplId=1146099689490845704, title=科学技术与工程, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1148622798802673703, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=articleTextType, value=kx, createTime=1751782615614, updateTime=1751782615614, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798781702180, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=banner, value=null, createTime=1751782615609, updateTime=1751782615609, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798769119267, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=j86gbwi+p0Idkyl5SzIlmQ==, createTime=1751782615606, updateTime=1751782615606, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798794285094, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1751782615612, updateTime=1751782615612, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798790090789, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1751782615611, updateTime=1751782615611, creator=18614031015, updator=18614031015)]), Website(id=1155914124811976731, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123166801305609, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/kxjsygc/EN, language=EN, createTime=1753521003206, createBy=18614031015, updateTime=1753521003206, updateBy=18614031015, name=科学技术与工程-英文站点, tplId=1146101810881728533, title=Science Technology and Engineering, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1155914371227308235, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=articleTextType, value=kx, createTime=1753521061952, updateTime=1753521061952, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371210531016, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=banner, value=null, createTime=1753521061947, updateTime=1753521061947, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371202142407, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=j86gbwi+p0Idkyl5SzIlmQ==, createTime=1753521061945, updateTime=1753521061945, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371223113930, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1753521061950, updateTime=1753521061950, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371218919625, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1753521061949, updateTime=1753521061949, creator=18614031015, updator=18614031015)])], journalTitle=科学技术与工程, weixinUrl=null, journalUrl=null, iacademicId=null, status=0, seqNo=null, journalTitleEn=Science Technology and Engineering, journalPhotoCn=UKU/O7GSka5polgCTkbIIw==, journalPhotoEn=5hwlULoNwcbj3xUmVi9MAQ==, journalFirstLetter=S, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=null, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/kxjsygc/CN/10.12404/j.issn.1671-1815.2309705, detailUrlEn=https://castjournals.cast.org.cn/joweb/kxjsygc/EN/10.12404/j.issn.1671-1815.2309705, pdfUrlCn=https://castjournals.cast.org.cn/joweb/kxjsygc/CN/PDF/10.12404/j.issn.1671-1815.2309705, pdfUrlEn=https://castjournals.cast.org.cn/joweb/kxjsygc/EN/PDF/10.12404/j.issn.1671-1815.2309705, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于自适应高斯混合模型与ResDN的火焰检测算法
收藏切换
PDF下载
王文标 , 时启衡 , 郝友维
科学技术与工程 | 论文·自动化技术、计算机技术 2025,25(4): 1580-1586
收起
收藏切换
科学技术与工程 | 论文·自动化技术、计算机技术 2025, 25(4): 1580-1586
基于自适应高斯混合模型与ResDN的火焰检测算法
全屏
王文标 , 时启衡, 郝友维
作者信息
  • 大连海事大学船舶电气工程学院, 大连 116026
  • 王文标(1982—),男,汉族,天津人,博士,讲师。研究方向:工业过程控制、机器视觉、图像分析与处理。E-mail:

Fire Detection Algorithm Combining Adaptive Gaussian Mixture Model and ResDN
Wen-biao WANG , Qi-heng SHI, You-wei HAO
Affiliations
  • School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China
出版时间: 2025-02-08 doi: 10.12404/j.issn.1671-1815.2309705
文章导航
收藏切换

针对火焰检测算法在复杂场景下误检率高、算法适应性差、效率低等问题,设计一种轻量高效的两阶段视频火焰检测算法。第一阶段采用改进的自适应高斯混合模型(adaptive gaussian mixture model,AGMM)对视频图像序列进行快速背景建模,利用火焰的闪烁和涌动特性,提取出序列中的可疑候选区域。第二阶段使用残差深度归一化卷积神经网络 (residual deep normalization and convolutional neural network, ResDN)对可疑候选区域进行判别,并引入简化的残差块替换原有的卷积层进行轻量化设计,实现对火焰的检测与定位。相比于传统分类算法,所设计的两阶段视频火焰检测算法能够有效克服复杂场景下的环境干扰,准确快速地识别火焰,具有更高的检测率和适应性。

火焰检测  /  自适应高斯混合模型(AGMM)  /  残差深度归一化卷积神经网络 (ResDN)  /  机器视觉  /  深度学习

A lightweight and efficient two-stage video flame detection algorithm was designed to address issues of high false positive rates, poor adaptability, and low efficiency in complex scenes. In the first stage, an improved adaptive Gaussian mixture model (AGMM) was employed for rapid background modeling of video image sequences. Suspicious candidate regions were extracted from the sequences by leveraging the flickering and surging characteristics of flames. In the second stage, a residual deep normalization and convolutional neural network (ResDN) was used to discriminate these suspicious candidate regions. A simplified residual block was introduced to replace the original convolutional layers for a lightweight design, enabling accurate flame detection and localization. Compared with traditional classification algorithms, the proposed two-stage video flame detection algorithm effectively overcomes environmental interference in complex scenes, rapidly and accurately identifies flames, and demonstrates higher detection rates and adaptability.

flame detection  /  adaptive Gaussian mixture model(AGMM)  /  residual deep normalization and convolutional neural network(ResDN)  /  machine vision  /  deep learning
王文标, 时启衡, 郝友维. 基于自适应高斯混合模型与ResDN的火焰检测算法. 科学技术与工程, 2025 , 25 (4) : 1580 -1586 . DOI: 10.12404/j.issn.1671-1815.2309705
Wen-biao WANG, Qi-heng SHI, You-wei HAO. Fire Detection Algorithm Combining Adaptive Gaussian Mixture Model and ResDN[J]. Science Technology and Engineering, 2025 , 25 (4) : 1580 -1586 . DOI: 10.12404/j.issn.1671-1815.2309705
随着中国城镇化进程加快,火灾隐患及危险源呈现越来越复杂和多样化的趋势,对社会发展和居民生命财产安全构成巨大威胁[1-2]。近年来基于视频图像的目标检测技术不断发展,许多学者开始研究了基于人为设计特征的火焰检测算法[3-6]。宋宁等[7]提出一种基于运动检测和像素颜色特征相结合的火焰检测方法。席廷宇等[8]提出了一种基于多特征量对数回归模型的火焰快速识别算法。Wang等[9]提出一种新的基于多专家系统的室内火灾检测方法。Xie等[10]利用基于动态的运动闪烁特征和深度静态特征进行火灾检测。然而,在现实环境中,存在如工厂风扇等周期性运动物体、静止光源及反射光线、特殊颜色的装置、太阳光线等环境因素的影响,严重限制了火焰检测的精度。随着深度学习技术的快速发展,基于深度学习的视频火焰检测成为主流研究方向[11-12]。目前,主要有单阶段与两阶段两种方法。单阶段检测器注重速度和实时性,其中常用的单阶段检测器包括YOLO(you only look once)系列[13-15]。两阶段检测器则能够更精确地定位目标并提供更准确的分类结果。孙维亚等[16]使用高斯混合模型(Gaussian mixture model, GMM)来提取运动区域,然后利用卷积神经网络进行判别。然而,经典的高斯混合模型存在局限性。由于预先设定高斯分布数量,在面对复杂、快速变化场景时效果不佳。同时,采用EM(expectation-maximization algorithm)算法迭代参数需要较长时间,且模型对噪声和异常值敏感,导致检测效果与检测速度均无法满足要求。吕苗苗等[17]提出一种基于改进高斯混合模型的优化背景建模方法,不仅降低数据计算量,还提高检测速度。侯涛等[18]对传统的高斯混合模型进行改进,设计自适应的高斯分布个数和学习率,有效提高了轨道异物入侵目标检测的精度。李笑等[19]提出四帧间差分结合改进的混合高斯模型算法, 提高了运动目标检测的准确性和实时性。
火焰通常呈现闪烁、涌动、扭曲和脉动等运动状态,而类伪火物体如静止光源、反射光线等则不具有这种运动状态,现通过自适应高斯混合模型能够有效将此类物体吸收为背景模型的一部分,而将火焰作为前景目标提取出来,随后将提取出的前景目标送入ResDN网络中进行判别,有效降低了误检率与漏检率,提高了火焰检测的可靠性与准确性。通过改进高斯混合模型并结合深度学习技术,能够有效地识别火焰,从而为城市安全管理提供更可靠的技术支持,减少火灾对居民生命财产的威胁。
高斯混合模型通过对每个像素点的时间序列数据进行建模来区分背景和前景,每个像素点Xt都使用一个高斯混合模型来表示,其函数表达式为
$\left\{\begin{array}{l}P\left({X}_{t}\right)=\sum _{i=1}^{K}{\omega }_{i,t}\eta ({X}_{t},{\mu }_{i,t},{\sigma }_{i,t},\lambda)\\ \eta ({X}_{t},{\mu }_{i,t},{\sigma }_{i,t},\lambda)=\prod _{\lambda =1}^{3}\frac{1}{\sqrt{2\pi \sigma }}{e}^{-\frac{1}{2}\left[\frac{{(x-\mu)}^{2}}{{\sigma }^{2}}\right]}\end{array}\right.$
式(1)中:P(Xt)为时间序列t下由多个高斯分布组成的概率密度函数;$\eta ({X}_{t},{\mu }_{i,t},{\sigma }_{i,t},\lambda)$为单个高斯分布的概率密度函数,其中,${\mu }_{i,t}$t时刻第i个高斯分布的均值;${\sigma }_{i,t}$t时刻第i个高斯分布的方差;K为高斯分布个数,K值越大,模型能表示的背景越复杂,同时计算量也会相应提升;σ为高斯分布的方差;μ为高斯分布的均值;$\lambda $为输入图片的RGB通道。
AGMM使用在线学习机制对原模型中每个高斯分布的权重、均值和方差进行更新,更新公式为
$\left\{\begin{array}{l}{\mu }_{i,t}=(1-\beta){\mu }_{i,t-1}+\beta {X}_{t,\lambda }\\ {\sigma }_{i,t}=(1-\beta){\sigma }_{i,t-1}+\beta ({X}_{t,\lambda }-{\mu }_{i,t}{)}^{2}\\ {\omega }_{i,t}=(1-\alpha){\omega }_{i,t-1}+\beta \\ \beta =P\left({X}_{t}\right)\alpha \end{array}\right.$
式(2)中:$\alpha $为学习率;${\omega }_{i,t}$t时刻第i个高斯分布所对应的权重;$\beta $为参数更新率,αβ共同决定背景模型更新的速度。
如果新像素点与现有模型的拟合程度不佳,则依照式(3),决定是否增加一个新的高斯分布。
$\left\{\begin{array}{l}D={X}_{t,\lambda }-{\mu }_{i,t}\\ P=D2.8\sqrt{{\sigma }_{i,t}}\\ M={P}_{\lambda =0}\bigcap {P}_{\lambda =1}\bigcap {P}_{\lambda =2}\end{array}\right.$
式(3)中:D为像素值与每个高斯分布的均值之间的差异;Xt,λ为针对一个视频序列中,t时刻的某通道上的像素值;P为像素值是否满足各个高斯分布的阈值条件;M为输入图片的RGB通道是否从属于同一个高斯分布,只有当RGB的3个通道都属于同一个高斯分布,才能认定该像素属于此高斯分布。
当不匹配的像素数量超过设定的增加阈值,就需要增加一个新的高斯分布;同理,若不匹配的像素数量低于设定的删除阈值,就需要删除权重最低的高斯分布。
动态调整高斯分布数量可以帮助模型更好地应对光照条件或天气因素的变化,适应不同环境条件,提高算法的鲁棒性和准确性。根据需要增加或减少高斯分布的数量减少不必要的计算开销,节省计算资源,提高背景更新的效率。
采用固定的学习率不能兼顾模型的适应性和稳定性要求,因此采用指数衰减变化的学习率策略。在模型建立初期,采用较大的学习率快速建立背景,随着背景模型趋于稳定,再根据状态动态调整学习率,实现自适应修正。保证模型以一个较快的速度学习数据特征,同时能够更稳定地收敛。
指数衰减学习率的计算公式为
${\alpha }_{t}={\alpha }_{t-1}{e}^{-bx}$
式(4)中:x为训练步幅;b为下降幅度;αtt时刻学习率;${{{\alpha }_{t}}_{-}}_{1}$为上一时刻学习率。
K个高斯分布按照${\xi }_{i,t}$的大小由高到低排序,取前B个高斯分布联合生成背景,随后将当前像素值XtK个高斯分布按优先级照式(5)进行匹配。
$\left\{\begin{array}{l}{\xi }_{i,t}=\frac{{\omega }_{i,t}}{{\sigma }_{i,t}}\\ {X}_{background}=\sum _{i=1}^{B}{\omega }_{i,t}{\mu }_{i,t}\\ B=\underset{b}{argmin}\left(\sum _{i=1}^{B}{\omega }_{i}T\right)\\ \left|{x}_{i,t}-{\mu }_{i,t}\right|2.5{\sigma }_{i,t}\end{array}\right.$
式(5)中:ξi,t为某个高斯分布中其权值与其方差的比值;Xbackground为最终创建的联合背景;ωi为权值;T为背景选取的阈值,T决定背景中高斯分布的个数,T值越大,选取的高斯分布个数越多。
XtK个高斯分布中属于背景中的某个高斯分布匹配,则为背景;否则该像素点被检测为前景,即运动目标像素。自适应高斯混合模型目标提取流程如图1所示。
深度归一化卷积神经网络(deep normalization and convolutional neural network, DNCNN)在处理烟雾这种不定形体时展现出了卓越的性能[20]。在此基础之上,根据火焰的特征对其结构进行改进,结构如图2所示。使用该网络训练模型,将一阶段中提取到的前景目标作为网络的输入,经过卷积层与池化层对内部特征进行提取,最后经由全连接层对目标进行分类。
ResDN网络是由14层深度归一化卷积神经网络(DNCNN)改进而来,通过将其中的一些卷积层替换为简化的残差块来解决深层网络中的梯度消失和梯度爆炸问题,以便更好地学习和传播信息。具体的参数配置如表1所示。
(1)输入层。网络的输入为48×48的3通道RGB图像。
(2)DN层。DN层由卷积层,批量归一化层,激活函数层组成,其中参数如上表所示,使用same模式进行Padding,卷积后叠加偏置作为激活的阈值,得到输出特征图上每一个元素vi,j,可表示为
${v}_{i,j}=f\left\{BN\right[conv2d(W,X)-b\text{'}\left]\right\}$
式(6)中:f为激活函数;conv2d(W,X)为二维卷积操作;W为卷积核;X为输入特征图;b'为偏置项;BN为批量归一化操作。
此过程实现了特征提取和非线性激活,使神经网络能够捕捉图像数据中的复杂特征。卷积层的输出特征图一般具有不同尺度,这会降低网络训练的稳定性。因此需要对特征图进行归一化处理,转换到统一的尺度下。使得各层之间的信息传递更加平衡和稳定,同时还可以加速训练过程,计算公式为
$y=\frac{x-mean\left(x\right)}{\sqrt{var\left(x\right)}-\epsilon }\lambda \text{'}+\beta \text{'}$
式(7)中:$\lambda \text{'}$β'为可学习参数;mean为均值;var为方差;$\epsilon $为用于防止分母为0的极小值,使用激活函数来增加网络的非线性。
经过卷积、归一化和激活操作后,神经网络可以更有效地提取和处理输入图像数据中的特征信息。
(3)简化的残差块。残差块对于改进深度神经网络性能发挥重要作用。其能够解决深层网络中的梯度消失和梯度爆炸等问题,能更好地学习和传播信息。常规的卷积神经网络中,信息会通过多个卷积层传递,但在传递过程中可能会出现信息丢失或衰减的问题。这会导致难以训练更深的网络,并且可能会影响网络的性能。引入残差块后,能够保留和传递更多的信息,减轻梯度在深层网络中的衰减问题,使得网络更容易训练和优化。其结构如图3所示。
(4)池化层。池化层用于降低参数与计算量,保留主要特征。MaxPooling1采用3×3步长为2的最大池化层,Padding模式为same,MaxPooling2和MaxPooling3使用2×2步长为2的最大池化层,Padding模式为valid。
(5)全连接层。全连接层的作用是对提取到的特征进行整合和映射。借助全连接层,网络可以在更高层次上理解输入数据,并对其进行更高级别的推理和分类。
实验使用的计算机CPU型号为12th Gen Intel(R) Core(TM) i5-12490F 3.00 GHz,内存为16 GB,显卡采用NIVIDIA GeForce RTX 3060,第一阶段自适应高斯混合模型使用Python3.8与Opencv-python4.5编写,第二阶段DNCNN网络使用tensorflow-gpu2.7.0版本搭建,cuDNN版本为8.2.1,CUDA版本11.7。
2.1 数据集准备
实验采用的数据集包含2.1×104张图像样本,图像格式为48×48×3。数据集中包含火焰图片10 269张,非火焰图片10 325张。这些图像样本是从工业现场实景拍摄、Kaggle等数据集中获取的。火焰视频数据共计41个,分别来自网络、韩国启明大学实验室[21]、土耳其比尔肯大学实验室/视觉研究室、美国NIST火灾研究实验室美国国家标准与技术研究院(National Institute of Standards and Technology,NIST)的火灾研究实验室(Fire Research Laboratory)以及研究者自行采集和制作的视频。
训练使用随机梯度下降法(stochastic gradient descent,SGD),学习率设定为0.01,以0.000 1的倍率进行学习率衰减,学习率衰减的计算公式为
${l}_{r,t}={l}_{r,t-1}\frac{1}{{e}_{pocℎ}d+1}$
式(8)中:lr,tt时刻学习率;${{{{L}_{r,}}_{t}}_{-}}_{1}$为上一时刻学习率;epoch为训练轮数;d为衰减倍率,为避免落入局部最优解,动量设定为0.9。
高斯分布数量K一般设置为3~5个,K初始值设置为3;初始学习率$\alpha $设置为0.05;背景阈值T设置为0.5;不匹配像素阈值设置为0.05;删除阈值设置为0.01;下降幅度参数b设置为0.2;均值(means)的初始值在0~255中随机选取;方差(variance)初始值在0~255中随机选取;权重系数(weights)初始值设置为$\frac{1}{k}。$
图4展示了不同算法进行建模和差分的效果。
选择多个视频样本来全面评估这些算法的性能。可以看出,在视频背景较为简单且色彩差距较大的情况下,HSV(hue, saturation, value)模型能够相对精确地提取火焰图像。然而,为了确保提取的准确性,每次提取前都需要提前设定准确的阈值,在实际应用中,这种方式既困难又不可取。GMM模型由于传统的EM算法导致其迭代速度慢,因此在轮廓检测时会提取出大量虚假目标,而AGMM模型使用在线学习机制,迅速完成对背景的更新,能提取出有较为干净的前景。
表2中统计了不同场景下使用HSV、GMM、本文方法建模后得到的样本检测率,每个场景均提取500张前景图片,统计其中火焰样本数量。
在Video4和Video5中,由于干扰因素较少,3种方法均表现出较高的检测率。而在Video1、Video2、Video3中,自适应高斯混合模型的性能明显优于HSV模型和传统的高斯混合模型,HSV模型由于受到如光照变化、阴影和反射以及颜色相近的静态干扰物等的影响,降低了检测率。传统的高斯混合模型由于高斯分布数量固定,不能很好地适应复杂背景。相比之下,自适应高斯混合模型能够快速更新参数,将周期性运动物体、反射光线、静止光源等吸收进背景,在前景提取任务中表现优异,在自然背景下的目标提取效果显著优于其他两种方法,具有更强的实用性和适应性。
图5为各网络训练过程中准确率(ACC Rate)、检出率(DR Rate)、虚警率(FAR Rate)和损失函数(LOSS)的变化趋势。
图5可以看出,训练集与验证集各有样本10 000张且已均衡化处理,每轮训练随机抽取36张,共1 000轮,其中在第150轮时各曲线都基本趋于稳定,训练过程基本达到预期。
使用自适应高斯混合模型对视频集进行处理,并将处理后得到的正负样本进行数据增强,形成测试集。此后使用本文方法和经典的卷积神经网络(convolutional neural network,CNN)网络(如VGG16、Res18和Res50等)对测试集进行对比试验,并将结果整理在表3中。
ResDN使用简化的残差块替换卷积层后,能在提高准确率、检出率,降低虚警率的同时降低其模型大小,与经典的CNN网络对比,也有较为明显的提升。
为解决工业现场火焰检测存在的适应性差,准确率低等,提出一种自适应高斯混合模型与ResDN相结合的视频火焰检测算法。该算法采用两阶段检测策略,一阶段使用自适应高斯混合模型对背景进行快速建模并提取出前景目标,自适应高斯混合模型能够对高斯模型的数量进行动态调整,从而适应不同时间下光照强度与环境的变化,提高背景模型精度与稳定性,解决传统模型在复杂环境下易受光照强度变化影响的问题;二阶段将前景目标送入ResDN网络中进行火焰识别,针对不同形态和大小的火焰均具有较好的识别效果。实验结果表明,所提出的二阶段检测算法能够适应多种复杂背景,有效降低提高火焰检测的准确率,降低虚警率。
  • 国家自然科学基金(52071047)
  • 国家自然科学基金(62073054)
参考文献 引证文献
排序方式:
[1]
陈志芬, 黄靖玲, 李亚. 适应城市消防规划需求的火灾风险评估研究[J]. 中国安全生产科学技术, 2019, 15(5): 185-191.
Chen Zhifen, Huang Jingling, Li Ya. Study on fire risk assessment adapting to requirements of urban fire protection planning[J]. Journal of Safety Science and Technology, 2019, 15(5): 185-191.
[2]
唐声强, 黄晓家, 谢水波, 等. 火灾对我国家庭财产保险需求影响的实证研究[J]. 消防科学与技术, 2023, 42(3): 416-421.
Tang Shengqiang, Huang Xiaojia, Xie Shuibo, et al. An empirical study on the impact of fire on the demand of family property insurance in China[J]. Fire Science and Technology, 2023, 42(3): 416-421.
[3]
侯逸臣, 何建, 王琳, 等. 基于多分类器融合的林火检测方法研究[J]. 计算技术与自动化, 2023, 42(1): 53-57.
Hou Yichen, He Jian, Wang Lin, et al. Research on forest fire detection method based on multi-classifier fusion[J]. Computing Technology and Automation, 2023, 42(1): 53-57.
[4]
卞永明, 高飞, 李梦如, 等. 结合K-means++聚类和颜色几何特征的火焰检测方法[J]. 中国工程机械学报, 2020, 18(1): 1-6.
Bian Yongming, Gao Fei, Li Mengru, et al. Fire detection method using K-means++ clustering and features of mixed color and geometry[J]. Chinese Journal of Construction Machinery, 2020, 18(1): 1-6.
[5]
葛珺, 王丽, 李伟, 等. 基于可见光图像的火焰识别与提取[J]. 消防科学与技术, 2020, 39(2): 247-249, 266.
Ge Jun, Wang Li, Li Wei, et al. Recognition and extraction of flames based on visible light images[J]. Fire Science and Technology, 2020, 39(2): 247-251, 266.
[6]
曹江涛, 秦跃雁, 姬晓飞. 基于视频的火焰检测算法综述[J]. 数据采集与处理, 2020, 35(1): 35-52.
Cao Jiangtao, Qin Yueyan, Ji Xiaofei. Review on video based flame detection algorithm[J]. Journal of Data Acquisition & Processing, 2020, 35(1): 35-52.
[7]
宋宁, 强彦, 董林佳. 林火监测中基于视觉的火焰检测方法[J]. 科学技术与工程, 2017, 17(25): 268-273.
Song Ning, Qiang Yan, Dong Linjia. Vision based flame detection method in forest fire monitoring[J]. Science Technology and Engineering, 2017, 17(25): 268-273.
[8]
席廷宇, 邱选兵, 孙冬远, 等. 多特征量对数回归的火焰快速识别算法[J]. 计算机应用, 2017, 37(7): 1989-1993.
Xi Tingyu, Qiu Xuanbing, Sun Dongyuan, et al. Fast fire flame recognition algorithm based on multi-feature logarithmic regression[J]. Journal of Computer Applications, 2017, 37(7): 1989-1993.
[9]
Wang T, Bu L, Yang Z, et al. A new fire detection method using a multi-expert system based on color dispersion, similarity and centroid motion in indoor environment[J]. IEEE/CAA Journal of Automatica Sinica, 2020, 7(1): 263-275.
[10]
Xie Y K, Zhu J, Cao Y G, et al. Efficient video fire detection exploiting motion-flicker-based dynamic features and deep static features[J]. IEEE Access, 2020, 8: 81904-81917.
[11]
吴雪, 宋晓茹, 高嵩, 等. 基于数据增强的卷积神经网络火灾识别[J]. 科学技术与工程, 2020, 20(3): 1113-1117.
Wu Xue, Song Xiaoru, Gao Song, et al. Convolution neural network based on data enhancement for fire identification[J]. Science Technology and Engineering, 2020, 20(3): 1113-1117.
[12]
杨剑, 刘方涛, 张涛, 等. 基于改进型残差网络烟雾图像识别[J]. 科学技术与工程, 2019, 19(32): 236-243.
Yang Jian, Liu fangtao, Zhang Tao, et al. Research on smoke image recognition based on improved residual network[J]. Science Technology and Engineering, 2019, 19(32): 236-243.
[13]
张旭彤, 胡鹏, 赵鑫, 等. 火灾预警中基于YOLOv5的火源智能检测定位方法[J]. 微电子学与计算机, 2023, 40(3): 67-74.
Zhang Xutong, Hu Peng, Zhao Xin, et al. Intelligent detection and location method of fire source based on YOLOv5 in fire warning[J]. Microelectronics & Computer, 2023, 40(3): 67-74.
[14]
陈露萌, 曹彦彦, 黄民, 等. 基于改进YOLOv5的火焰检测方法[J]. 计算机工程, 2023, 49(8): 291-301, 309.
Chen Lumeng, Cao Yanyan, Huang Min, et al. Flame Detection Method Based on Improved YOLOv5[J]. Computer Engineering, 2023, 49(8): 291-301, 309.
[15]
谢康康, 朱文忠, 肖顺兴, 等. 一种改进YOLOX_S的火焰烟雾检测算法[J]. 科学技术与工程, 2024, 24(8): 3298-3307.
Xie Kangkang, Zhu Wenzhong, Xiao Shunxing, et al. Flame smoke detection algorithm based on improved YOLOX_S[J]. Science Technology and Engineering, 2024, 24(8): 3298-3307.
[16]
孙维亚, 陈恺鑫, 吴铭, 等. 融合运动特征的高效视频火焰检测算法[J]. 数据采集与处理, 2021, 36(6): 1276-1285.
Sun Weiya, Chen kaixin, Wu Ming, et al. An efficient video flame detection algorithm integrating motion features[J]. Journal of Data Acquisition & Processing, 2021, 36(6): 1276-1285.
[17]
吕苗苗, 孙建明. 基于改进高斯混合模型的运动图像目标检测算法[J]. 半导体光电, 2019, 40(6): 874-878, 885.
Miaomiao, Sun Jianming. Moving image target detection based on modified Gaussian mixture model[J]. Semiconductor Optoelectronics, 2019, 40(6): 874-878, 885.
[18]
侯涛, 宝才文, 陈燕楠. 基于自适应高斯混合模型的铁轨异物入侵检测研究[J]. 光电子·激光, 2022, 33(4): 403-413.
Hou Tao, Bao Caiwen, Chen Yannan. Research on detection of foreign object intrusion in railroad tracks based on AGMM[J]. Journal of Optoelectronics·Laser, 2022, 33(4): 403-413.
[19]
李笑, 杨宇, 徐一鸣. 四帧间差分和改进混合高斯模型对运动目标的检测[J]. 科学技术与工程, 2020, 20(15): 6141-6150.
Li Xiao, Yang Yu, Xu Yiming. Detection of moving targets by four-frame difference and modified gaussian mixture model[J]. Science Technology and Engineering, 2020, 20(15): 6141-6150.
[20]
Yuan F N, Xia X, Shi J T, et al. Non-linear dimensionality reduction and gaussian process based classification method for smoke detection[J]. IEEE Access, 2017, 5: 6833-6841.
[21]
Deok-Yeon K, Joon-Young K, Chul K B, et al. Human detection using wavelet-based CS-LBP and a cascade of random forests[C]// IEEE International Conference on Multimedia and Expo. Melbourne: IEEE, 2012: 362-367.
2025年第25卷第4期
PDF下载
341
128
引用本文
BibTeX
文章信息
doi: 10.12404/j.issn.1671-1815.2309705
  • 接收时间:2023-12-08
  • 首发时间:2025-07-29
  • 出版时间:2025-02-08
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2023-12-08
  • 修回日期:2024-11-15
基金
国家自然科学基金(52071047)
国家自然科学基金(62073054)
作者信息
    大连海事大学船舶电气工程学院, 大连 116026
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/kxjsygc/CN/10.12404/j.issn.1671-1815.2309705
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏