A gait recognition based on link model of infrared thermal imaging

In the environment of complex background, load and night, the correct gait recognition rate is greatly affected in visible image, in order to solve this problem, this paper use thermal infrared imager to capture image in different body, different scene and different angle, then image threshold, frame-difference and contour auto adjustment method are used to extract human body outline, the feature vector of each joints are extracted by modified five link model, after that the extracted feature vector of each joint is passed to the SVM classifier to identify the characters. Leave-one-out method is used to calculate the correct rate of recognition, the final recognition correct rate is between 71-92%, in the night, load and other complex background scenes, the correct recognition rate is much better than visible light.