Pedestrian recognition, positioning and counting method for video

Abstract

The invention provides a pedestrian recognition, positioning and counting method for a video. The method achieves the recognition, positioning and counting of pedestrians in a video through the matching angle of the shape context of each point in a video image, and carries out the matching with sample points in each image frame of the inputted video to be recognized through building a standard feature library containing the shape context features of sample points, so there is no need to carry out a large amount of training for the standard feature library and only the standard feature library containing typical body postures needs to be built. Moreover, the method enables a body image to be segmented into the sample points even if a human body is complex. Because each sample point is matched with the corresponding sample point in the standard feature library, the method is higher in recognition precision, reduces the recognition error, is wider in application range, and can be further applied to the main body recognition and tracking in the video.

Description

technical field [0001] The invention relates to a person identification, location and statistics method, in particular to a person identification, location and statistics method in video. Background technique [0002] Person recognition is one of the important problems in computer image and video processing. The essence of person recognition in images and videos is to classify and label the subjects in the images according to the given preconditions given the training and test image sets. Effective target recognition can find important subjects in the image, including people or other objects to be recognized, so as to provide basis and clues for further recognition and tracking. Currently, there have been many literatures proposing and creating models for subject recognition in images or videos. For example, when identifying objects with texture features, commonly used texture feature operators can be used to classify objects. In other methods, given some fixed scenes, re...

AI Technical Summary

Problems solved by technology

Among them, the global feature and local feature methods require a large amount of training data due to passive learning, which increases the difficulty of recognition.

However, due to the innate complexity of the human body, the method of joint space distribution is limited to the identification of fewer parts.

The part library method needs to repeatedly observe the appearance of small parts of each subject in the space, which increases the amount of data and the complexity of processing

The deformable template method needs to make various templates that conform to the shape and posture of the human body in advance, while the applied graph segmentation method can only recognize fewer target objects at present

Images