Pedestrian re-identification method based on fuzzy robustness features and pedestrian re-identification module device based on fuzzy robustness features

Abstract

The present invention relates to a pedestrian re-identification method based on fuzzy robustness features. The method comprises the following steps of: (a) performing fuzzy processing of all the images in a disclosed large-scale face image database through a fuzzy kernel to obtain large-scale image data approaching pedestrian image quality; (b) employing a processed face image data to train a deepconvolutional neural network model; (c) performing parameter tuning of the pre-trained deep convolutional neural network model by employing a current small-scale pedestrian image database; and (d) performing feature extraction of pedestrian images, and performing image re-identification through feature comparison. The pedestrian re-identification method based on fuzzy robustness features and thepedestrian re-identification module device based on fuzzy robustness features are characterized in that: the pre-trained deep convolutional neural network model has fuzzy robustness for pedestrian images and a high generalization ability; and a good effect can be achieved for different scales and different types of pedestrian re-identification image databases through adoption of the method provided by the invention.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a pedestrian re-identification method and module device based on fuzzy robust features, which solve the problems of blurred images of pedestrians to be identified under a camera and a small number of training samples. Background technique [0002] With the development of science and technology, video surveillance systems have been widely used in communications, security, transportation and other industries. They play an important role in maintaining social order and are gradually becoming popular and intelligent. Currently, camera networks have covered most public places, such as airports, railway stations, subways, supermarkets and highways. The construction of public camera network is playing an increasingly important role in the practice of combating crime and maintaining law and order. Pedestrian re-identification technology is a research field based on computer visi...

AI Technical Summary

Problems solved by technology

[0003] Under actual conditions, the existing pedestrian re-verification database is small in scale, and each recognition target appears less frequently under the camera, resulting in less available training sample data, which makes the trained deep learning recognition model seriously overfit. combined problem

At the same time, the limitation of the quality of surveillance cameras and the movement of pedestrians also lead to low resolution and blurred images of pedestrians, seriously reducing the quality of pedestrian images, which can easily lead to misidentification of the final target

Images