Single-view-angle low-resolution gait feature extraction method and device and storage medium

A gait feature and low-resolution technology, which is applied in the field of single-view low-resolution gait feature extraction, devices and storage media, can solve the problem that the gait feature extraction method is difficult to meet the accurate description of gait, and achieve calculation Convenience, elimination of data redundancy, effect of maintaining information

CN112528844APending Publication Date: 2021-03-19CENT SOUTH UNIV
4 Cites 2 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2021-03-19

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a single-view-angle low-resolution gait feature extraction method and device and a storage medium. The method comprises the steps: acquiring a monitoring image; processing themonitoring image by adopting an interpolation-based super-resolution algorithm to obtain a high-resolution monitoring image; performing spatial local feature extraction on the high-resolution monitoring image by adopting a Gabor filter to obtain gait Gabor features; performing dimension reduction processing on the gait Gabor features based on a maximum mutual information amount method, so as to obtain initial gait features; and performing secondary dimension reduction and noise reduction processing on the initial gait feature based on a principal component analysis method to obtain a final gait feature. According to the method, the super-resolution technology is used for extracting the high-resolution information, the high-resolution information is filtered through the Gabor filter and then subjected to dimensionality reduction twice, the final gait feature is obtained, spatial feature expression under the low-resolution and unclear conditions can be obtained, and the problem of spatial gait feature expression when a single-view-angle fixed camera is switched to a multi-view-angle camera is solved.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates, in particular, to a single-view low-resolution gait feature extraction method, device and storage medium. Background technique

[0002] Gait recognition is currently one of the important means of long-distance biometric recognition, and has great application scenarios in security detection, criminal investigation, and public domains. Spatial feature extraction based on human body structure has the advantages of simple representation of feature information, relatively convenient extraction, and certain robustness to noise and loss. However, it is difficult to accurately describe gait by directly extracting feature information from human body shape. For example, in a low-resolution complex surveillance scene under a single viewing angle, it is difficult to accurately extract the spatial features of the human body structure from the surveillance image. This may cause different targets to have the same feature expression. ...

Examples

Embodiment Construction

[0069] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described in detail below. Apparently, the described embodiments are only some of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other implementations obtained by persons of ordinary skill in the art without making creative efforts fall within the protection scope of the present invention.

[0070] Such as figure 1 As shown, an embodiment of the present invention provides a single-view low-resolution gait feature extraction method, including:

[0071] S01: Acquiring surveillance images.

[0072] S02: Process the surveillance image with a super-resolution algorithm based on interpolation to obtain a high-resolution surveillance image. Specifically include:

[0073] The monitoring image is interpolated by the bicubic convolution interpolation meth...