Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Pedestrian r re-identification method based on color matching

A technology for pedestrian re-recognition and color matching, used in character and pattern recognition, instruments, computer parts, etc.

Inactive Publication Date: 2019-01-25
THE 28TH RES INST OF CHINA ELECTRONICS TECH GROUP CORP
View PDF4 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Therefore, by extracting the appearance color information of the pedestrian's torso and legs, it is possible to effectively search for relevant target pedestrians in the video, but the existing technology has not yet involved research in this area, so a new technical solution is needed to achieve a breakthrough

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Pedestrian r re-identification method based on color matching
  • Pedestrian r re-identification method based on color matching
  • Pedestrian r re-identification method based on color matching

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0067] figure 1 It is a schematic diagram of the pedestrian re-identification system provided by Embodiment 1 of the present invention, refer to figure 1 , the specific method steps of the present embodiment are as follows:

[0068] 1) Detect the motion area in the video to get the pedestrian candidate area:

[0069] figure 2 is to determine the region of motion in the video, refer to figure 2 , forming a candidate area for pedestrian detection, including

[0070] 1.1) ViBe background modeling algorithm is used to detect the motion area in the video, and the foreground area of ​​each frame picture is obtained;

[0071] 1.2) The foreground area is processed by means of dilation and erosion to obtain motion zoning, thereby forming pedestrian candidate areas.

[0072] 2) For the obtained pedestrian candidate area, use the head and shoulders detector to further obtain the preliminary pedestrian area, and then use the DPM model to perform fine pedestrian detection:

[0073]...

Embodiment 2

[0124] In order to verify the performance of the color feature descriptor obtained by the look-up table quantization method, experiments are carried out. The experimental data are images of pedestrians collected in the real environment. These images come from surveillance videos of different fields of view, at different times and under different lighting conditions. There are about 5,000 images of pedestrians, from more than ten monitoring points, and the colors of pedestrians' clothes are different. .

[0125] Pedestrian pictures of the same type of color are calculated in pairs, similarities are calculated in pairs of pedestrians of different colors (using the intersection of the histogram in Embodiment 5 to calculate the vector similarity), obtain as follows Figure 6 The similarity threshold distribution plot shown. The dotted part on the right side represents the similarity distribution of the same color, and the oblique part on the left represents the similarity distrib...

Embodiment 3

[0127] Using color descriptors as features, retrieve a certain color in the data set and rank them by similarity. The ideal retrieval effect is that objects of the same color are ranked first, and objects of different colors are ranked lower. In this embodiment, the average accuracy rate mAP is used as a measure index, and the experimental results are shown in Table 2, which is the average accuracy rate of each color provided in this embodiment.

[0128]

[0129] Table 2.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a pedestrian re-identification method based on color matching, comprising the following steps: a moving area in a video is detected to obtain a pedestrian candidate area; the preliminary pedestrian area is obtained by using head-shoulder detector for pedestrian candidate area, and the DPM model is used for pedestrian precision inspection, by tracking the detected pedestrians, the target tracking sequence is obtained, the pedestrian area is divided into torso and legs, the color descriptor of pedestrian is obtained by using the proposed color quantization method, and thetarget sequence information is stored in the database. The histogram intersection method is used to calculate the similarity between the query targets and the pedestrian sequences in the database, and the query results are returned in descending order of similarity. The method of the invention can stably detect and track pedestrians in different background environments and under different monitoring cameras, and the proposed color quantization method can effectively cope with the influence of different pedestrian postures, resolutions and illumination, and realize cross-camera searching and querying the target sequence similar to pedestrians.

Description

technical field [0001] The invention belongs to the technical field of intelligent surveillance video, and in particular relates to a cross-camera pedestrian re-identification method based on color matching. Background technique [0002] At present, more and more attention is paid to the intelligent analysis of targets in surveillance video. For most monitoring systems, the people appearing in the monitoring video are the important objects of attention. Therefore, the intelligent monitoring system needs to have the ability to detect, track and identify them, so as to further analyze the behavior of pedestrian targets. Re-identification technology is one of its core steps. Pedestrian re-identification is to re-identify the same pedestrian that appears in different cameras. It is mainly used in multi-camera video surveillance networks in large areas, and can realize cross-camera trajectory tracking and trajectory prediction of pedestrian targets. Through the analysis of the ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/00G06K9/46
CPCG06V40/103G06V10/50G06V10/56
Inventor 李晓冬贺成龙吴蔚徐建平王鑫鹏
Owner THE 28TH RES INST OF CHINA ELECTRONICS TECH GROUP CORP
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com