Single target tracking method based on feature compensation

Abstract

The invention discloses a feature compensation video target tracking method based on a posterior pixel color histogram, a direction gradient histogram and a convolutional neural network, which uses simple features in a simple scene to ensure real-time performance, and uses complex features in a complex scene to ensure accuracy. Through combination of two features of a posterior pixel histogram anda direction gradient histogram, an obtained response feature map can well adapt to a simple condition of a video scene. Training a classifier to judge which situation the response obtained by the former is fused is untrusted; and finally, according to the judgment result of the classifier, selecting whether to add a convolutional neural network tracker which is slow in speed and more robust in performance to correct the target deviated from the tracking, or re-finding the target lost from the tracking. According to the method, the precision of judging the size and the position of the target in the video is improved, and the method can well adapt to a long-time target tracking task to achieve a scene of practical application.

Description

technical field [0001] The invention belongs to the technical field of single target tracking of computer vision, in particular to a single target tracking method based on feature compensation. Background technique [0002] In the field of computer vision, tracking tasks have always been a core problem, and are widely used in many aspects such as video surveillance, human-computer interaction, robot visual perception, and military guidance. Single target tracking is to manually mark the position and size of the tracking target with a rectangular frame in the first frame of the video, and then what the tracking method needs to do is to follow the manual mark with a rectangular frame in the subsequent frames of the video. objects. Similar to target detection, it is to scan and search for targets in the entire frame range of static images or dynamic videos. In a nutshell, target detection focuses on positioning and classification. Target tracking, on the other hand, focuses o...

AI Technical Summary

Problems solved by technology

[0004](1) Poor robustness

[0005] After the tracking model loses the target, it cannot be retrieved. This type of algorithm is mainly in the context of the accuracy of the position and size of the tracking target, and does not have a high Robustness, unable to adapt to long-term tracking tasks, such a model cannot be used in real-world scenarios very well

[0006] (2) Slow speed

[0007]Whether it is an end-to-end neural network structure tracking model or a tracking model combining deep convolutional feature maps and correlation filtering, although it can get a higher accuracy rate, but It takes a lot of computing time, so it is rarely used in actual scenarios

However, other traditional tracking models based on correlation filtering can achieve very fast speeds, but they are not good enough in terms of accuracy and robustness.

[0008](3) Error accumulation

Images