Pedestrian re-identification method and device based on space-time analysis and depth features

Abstract

The invention discloses a pedestrian re-identification method and device based on space-time analysis and depth features. In pedestrian re-identification application, global search for pedestrian images in an actual large-scale video monitoring scene has complexity and irrationality, and in order to further improve the identification accuracy and the identification speed, the invention provides amethod combining spatio-temporal information analysis and depth feature extraction. Firstly, the moving speed of a pedestrian is obtained through analysis to accord with gamma distribution, and then the space-time information of the pedestrian is further analyzed through the distribution to obtain the space-time prior probability of the pedestrian; training a convolutional neural network on the large-scale data set in combination with an actually acquired image, and extracting depth features to calculate a visual space-time probability; and finally, combining the two probabilities to judge whether the two images are the same pedestrian or not. The pedestrian re-identification efficiency can be effectively improved from massive monitoring or collected data in an actual large-scale video monitoring application scene, the high pedestrian re-identification precision can be kept, and the efficient and accurate pedestrian re-identification effect is achieved.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a pedestrian re-identification method and device based on spatio-temporal analysis and depth features. Background technique [0002] Pedestrian re-identification is also called pedestrian re-identification. It is to recognize pedestrian images taken at different times under the condition of cross-camera and cross-scene, so as to judge whether they are the same pedestrian, that is, to identify whether the pedestrians in cross-camera and cross-scene are the same person. With the continuous improvement of monitoring networks in society, and the growth of demand in smart cities, smart security, smart monitoring and other fields, research on pedestrian re-identification has attracted many researchers from industry and academia. [0003] Although face recognition technology has developed relatively maturely and has been applied in many scenarios and products, the application of face reco...

AI Technical Summary

Problems solved by technology

[0003] Although face recognition technology has developed relatively maturely and has been applied in many scenarios and products, the application of face recognition technology has certain limitations: First, face recognition technology can only be used for people with a human body. Face information, other important information is not fully utilized, such as: clothing, posture, behavior, etc.; secondly, face recognition technology must have a clear frontal face photo when it is applied, that is, the image details are relatively high. These conditions cannot be met in many scenarios, such as: lowering the head and side face, facing away from the camera, blurring the body shape, blocking the hat, etc.

Image-based pedestrian re-identification generally only focuses on the image information of pedestrians in actual scenes. In practical applications, such as in video surveillance, urban supervision, and criminal security, pedestrian images are often obtained from images that are far away from the target under cross-camera and cross-scenario situations. There are differences between different camera devices, and the details of pedestrians in the collected images are not obvious. Due to changes in clothing or posture, different categories of pedestrian images may be similar, but images of the same type of pedestrians are not. Very similar, it is difficult to identify; on the other hand, due to the large flow of pedestrians, the long monitoring time span, and the complex background, the obtained monitoring data is often massive, making the collected pedestrian image data very large; these practical difficulties lead to If a complex algorithm based on pedestrian images is used to improve the recognition accuracy of pedestrian re-identification, it will inevitably face the problems of complex calculations and increased system operating costs, which will directly lead to low recognition efficiency and increased system costs; The speed index is often accompanied by the decline of system recognition accuracy

Images