A method of identity recognition based on gait images

Abstract

The invention discloses an identity recognition method based on a gait image, which includes a training step of performing pedestrian detection and image preprocessing on the gait image sequence and assigning a label value to the corresponding gait image, and then using it by convolution receiving The feature learning network composed of limited Boltzmann machine and fully connected layer is trained to generate a feature learning network model and feature center value model with identity recognition; the identification step is to perform pedestrian detection and image preprocessing on the gait image to be identified, The gait periodicity is calculated by the normalized autocorrelation function to obtain a periodic gait sequence, and the identity of pedestrians is identified through a deep learning network and a voting algorithm. The present invention takes periodic gait image sequence as input, retains complete gait information; adopts deep learning network for feature learning, obtains more distinguishing gait features to improve recognition rate; through deep learning network and voting Algorithms are used together to enhance the accuracy and robustness of recognition.

Description

technical field [0001] The invention relates to the technical field of identification, in particular to an identification method based on gait images. Background technique [0002] Public security is related to social stability, economic development and other major issues, and has long attracted people's attention. An important part of this is fast and accurate identity authentication for public safety. Therefore, related identification technologies have also emerged as the times require. [0003] At present, commonly used identification technologies based on biological characteristics include iris recognition, face recognition and fingerprint recognition. These identification technologies have brought great convenience and security to daily life, such as mobile payment, face attendance, etc. . Although these identification technologies have brought many conveniences to people, they are also easily affected by factors such as distance, image resolution, and illumination, ...

AI Technical Summary

Problems solved by technology

[0008] The identification based on gait images can solve some disadvantages of the current common identification methods, but the existing gait recognition methods have problems such as manual design of features, long training time, low recognition rate, and poor robustness.

Images