Accurate and rapid recognition is often required for the king oyster mushroom under dark and humid environments using machine vision. Particularly, the mushrooms can intertwine with mycelial networks after imaging. This study aims to accurately locate king oyster mushrooms under dark and damp environments. An identification framework of the king oyster mushroom was proposed using an infrared array. A test bench was established based on an infrared reflection module. Sensing distance and sensing space cross-sectional radius of the infrared reflection module were then measured under different electrical parameters. A numerical simulation was used to fit the relationship between electrical parameters and infrared sensing distances. A mathematical model was obtained for the infrared reflection under typical parameter conditions. The invisible infrared light was then visualized for mushroom identification, according to the sensing space model from the infrared reflection module. A detection gantry (including multiple linearly arranged infrared reflection modules) was placed over a tray. The tray moved at a constant speed. The output levels of the infrared reflection modules were sampled periodically by an STM32F103ZET6 controller. An information matrix was formed for the position and shape of the mushrooms within the tray area. The sensing information matrix was determined by the number of timed samplings and the infrared reflection modules. Simple matrix operations were performed on the single-connected regions with zero values in the sensing information matrix. According to the morphologies of the king oyster mushroom, a double-cylinder rotary harvester was designed to calculate the position coordinates of the mushroom's center point relative to the tray. A servo motor drove a reducer, thus causing two coaxially arranged hollow cylinders to rotate relative to each other. Six blades evenly distributed below the cylinders were used to cut the king oyster mushroom in a planar motion. The harvester was fixed to a vertically moving linear module. The motion axes were arranged on an execution gantry, thereby driving the horizontal and vertical movements of the harvester. A king oyster mushroom harvesting device was developed, where the detection and execution gantries were sequentially placed on the tray. Taking king oyster mushrooms as the targets, five symmetrically distributed feature positions on the tray were selected for the identification and harvesting experiments. A series of experiments was conducted to verify the device. The time to solve for the mushroom center point coordinates was 3.5-4.5 ms using the information matrix, which was significantly less than the 40-50 ms required for mushroom image processing using machine vision. The high efficiency of the identification was obtained using the infrared array. Meanwhile, the deviation between the inner hole of the harvester and the mushroom cap center was controlled within 1.5-4.0 mm after the harvester was positioned on the mushroom. This gap met the operational quality requirements of mushroom harvesting robots. The harvesting success rate reached 100%, indicating the high positioning accuracy of the motion axis. The king oyster mushroom identification was developed using typical infrared reflection modules, whose cost was only 1/20 of the mainstream machine vision products. The outstanding efficiency and economic advantages were achieved in the strongly interfering substrate and mycelial networks. The findings can also provide a technical pathway for the edible king oyster mushroom harvesting under complex backgrounds.