Cross-view gait recognition method combining LSTM and CNN

Abstract

The invention discloses a cross-view gait recognition method combining LSTM and CNN, which comprises the following steps of: 1) extracting 3D (Three-Dimensional) attitude data of pedestrians in a video by adopting OpenPose3D from a CASIA-B gait video data set; 2) extracting motion constraint data and joint action data of video pedestrian joint points from the 3D attitude data obtained in the step1), constructing a pedestrian gait constraint matrix, and establishing a sample set; (3) constructing a 3D gait recognition network LC-POSEGAIT; 4) training a 3D gait recognition network LC-POSEGAIT by using the training set sample obtained in the step 2); and 5) utilizing the trained 3D gait recognition network LC-POSEGAIT to extract gait feature vectors of the video pedestrians, and completing cross-view gait recognition combining LSTM and CNN. The method can realize cross-view gait recognition, avoids preprocessing work of pedestrian detection and tracking, and can alleviate the influence of factors such as shooting angles, carried articles and illumination on gait recognition.

Description

technical field [0001] The invention belongs to the field of machine vision and relates to a cross-view gait recognition method combining LSTM and CNN. Background technique [0002] Biometric recognition has a wide range of applications in security monitoring, pedestrian tracking, identification and other fields. At present, there are a large number of camera monitoring equipment installed in public places, and the collected monitoring data is mostly used for evidence collection after the event, and seldom conducts identity verification for specific target pedestrians without perception, long distance, and uncontrolled conditions to support security warnings and Call the police. Gait is the performance of people's walking posture in the process of walking. It has the characteristics of non-contact, non-invasive, difficult to hide and forge. It is the only biological feature that can be obtained in a long-distance and uncontrolled state, and can support long-distance Identi...

AI Technical Summary

Problems solved by technology

However, human joint point tracking is a difficult point in traditional 3D gait models, and reconstruction of 3D gait models using 3D equipment or multi-camera collaboration requires complex camera parameter adjustment and modeling calculations, and its application scenarios are limited

Images