An infrared gait dual-channel feature fusion recognition method, the steps are: infrared image preprocessing, firstly using the background subtraction method of the mixed Gaussian model to realize the detection of moving human body; then performing morphological processing, that is, using morphological filtering to eliminate binary values Noise in the image and fill in the lack of moving targets; infrared image feature extraction, first of all, it is necessary to divide the cycle of the gait sequence, and determine the gait cycle based on the periodicity of the contour width; determine the gait energy map; infrared pyroelectric features Extraction, first of all, data collection, then denoising processing on the collected voltage signal, and then performing Fourier transform on the time domain signal to obtain its spectrum, and then use PCA to reduce the dimension to obtain the feature vector; classification and identification. In the present invention, the one-dimensional voltage signal collected by infrared pyroelectricity during human movement is easy to realize in real time, and the infrared image is not sensitive to the shape change of the carrying objects such as bags, umbrellas, backpacks, etc. when people walk, and it can also be used at night working.