Target vehicle extraction method based on video

Abstract

The invention discloses a target vehicle extraction method based on a video, and aims to solve the problem that result video compression accuracy is low since parts of frames of target vehicle fails in being extracted since a target vehicle extraction process ignores a problem that the feature information of the same vehicle has differences in different frames in the prior art. Through first-time feature extraction, a target vehicle is screened, a static vehicle is removed, then, a subsequent video frame and the target vehicle are subjected to similarity matching, feature extraction and matching does not need to be carried out if the similarity matching is successful, and the problem that feature information reflected by the same vehicle in different video frames has certain differences to cause feature extraction errors so as to affect compression accuracy is avoided. By use of the method disclosed by the invention, result video compression accuracy is improved, and meanwhile, the technical barrier that the prior art considers that the video frame needs to carry out feature extraction frame by frame is overcome. The method is suitable for the field of video image processing.

Description

technical field [0001] The invention relates to the field of video image processing, in particular to a method for extracting a target vehicle based on video. Background technique [0002] With the rapid development of society, the social economy has made great progress, and the traffic has changed with each passing day. Vehicles have greatly changed people's lives, the pace of society has been greatly accelerated, and people's work efficiency has been greatly improved. But vehicles also bring many social problems, such as frequent traffic accidents, so how to supervise the form of vehicles has become a hot spot. Currently, cameras are used for video surveillance, but the amount of video data is large and the content of the data is large, so how to effectively extract the target vehicle has become another problem. There are various target frame extraction methods such as automatic license plate recognition, loop coil detection method, and pre-divided recognition method. A...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to solve the problem that the target vehicle extraction process in the prior art ignores the difference in the feature information of the same vehicle in different frames, which leads to the failure of the target vehicle extraction in some frames, resulting in low accuracy of the resulting video compression. Provides a video-based target vehicle extraction method with higher compression accuracy

Images