A Gait Recognition Authentication Method Based on Feature Fusion

Abstract

A gait recognition authentication method based on feature fusion, the method includes: first, the camera collects the current background image and the gait original image sequence of the detection target in real time, and obtains the binary step by grayscale, European algorithm, median filter, etc. Then directly extract the pixel points of the person’s outline to obtain the static eigenvalue based on the aspect ratio and the dynamic eigenvalue based on the centroid distance of the gait outline; finally, the new algorithm combined with the SVM algorithm and the Bayesian algorithm is used to Detect target classification and output the final recognition result. This method can quickly and accurately remove the background, and improves the adaptability in different environments. In addition, because the prior probability of the traditional Bayesian algorithm is artificially set based on past experience, resulting in a low recognition rate, and the integration of the SVM algorithm can improve the prior probability of the Bayesian algorithm, making the overall The recognition accuracy can be significantly improved.

Description

technical field [0001] The invention relates to the technical field of gait recognition, in particular to a gait recognition authentication method based on feature fusion. Background technique [0002] Gait recognition is the use of computer technology to confirm identity through people's walking style and posture. Using gait as a biological feature to identify identity has some unique advantages: it can collect data from a long distance, and complete the identification without the knowledge of the subject. Gait recognition has become one of the hot spots in the field of biometric identification in recent years. From the research results at home and abroad, the difficulties of gait recognition mainly focus on background modeling, gait feature extraction, gait recognition speed and accuracy, etc. [0003] 1. In terms of background modeling, removing the background is a major problem. From the current research results, most of the methods for removing the background use the i...

AI Technical Summary

Problems solved by technology

These feature extraction methods are more complex

[0005] 3. In terms of gait recognition speed and accuracy, common gait recognition systems use low-efficiency algorithms and mostly rely on the support of supercomputers in the operating environment, which cannot meet actual needs on popular PCs

At present, there is no commercial gait recognition software. As far as the existing gait recognition programs are concerned, the recognition rate is not very ideal

Images