Multimodal feature fusion and classification recognition of infrared small targets

The detection of infrared small targets has important application value in military reconnaissance, security monitoring and other fields, but it faces challenges such as insufficient information and robustness due to the characteristics of small target size, low contrast and complex background. In this paper, the multi-source features of the target are used as the data basis, and complementary characteristics of the optical imaging characteristics, infrared radiation characteristics and motion characteristics of infrared weak small targets are used comprehensively. The multi-modal feature fusion and classification recognition model is proposed using the multi-model iterative mode, which can improve the recognition accuracy of infrared weak small targets, and achieve a high-efficiency match between recognition performance and speed in practical engineering.