Hyperspectral remote sensing technology has been widely used in many fields due to its high spectral resolution and rich spectral information. Object classification is one of the key techniques to fully use the hyperspectral data. Based on the investigation and summary of the research status of hyperspectral image classification technology，experiments were conducted in the mining area north of Jinchang，Gansu province. A comparative analysis was mainly carried out from two aspects：supervised classification and unsupervised classification. Taking the spectral Angle method as an example，the key factors affecting the classification performance of different methods were deeply discussed. The results show that the accuracy of supervised classification method is better than that of unsupervised classification for the areas with insufficient spatial characteristics like the experimental area, in which the Maximum Likelihood Classification is the best, and it also proves that unsupervised classification is not suitable for mineral classification in the similar areas.