Real-time correlation filtering tracking algorithm adaptive to scale change

Abstract

The invention discloses a real-time correlation filtering tracking algorithm adaptive to scale change, and the algorithm comprises the following steps: S1, obtaining a video sequence, setting the target position and scale of an initial frame, carrying out the sampling of a target, and carrying out the width-to-height ratio adjustment to initialize three filters; S2, reading a next frame of image,performing correlation filtering on the filter and the candidate area image to obtain a confidence map, and obtaining a target position; S3, carrying out sampling of three different scales at the target position according to each aspect ratio in a filter, zooming a sampling result and carrying out correlation filtering to acquire a confidence map of each scale, and acquiring a target scale; S4, adding the target position and the target scale into the training of the next frame, and performing model updating on the three filters; s5, judging whether the video is ended or not, and if so, ending;and if not, entering the step S2. According to the method, the influence caused by different aspect ratios is solved by initializing filters with different proportions, so that the purpose of adaptive scale estimation is achieved, and the tracking effect is improved.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a real-time correlation filter tracking algorithm adaptive to scale changes. Background technique [0002] Object tracking is the focus and one of the most challenging basic tasks in the field of computer vision. Its purpose is to locate the object of interest in the image sequence and track it continuously, stably and accurately. The correlation filter tracking algorithm builds a classifier between the target and the background by using the circular matrix theory and minimizing the error model. The tracking effect of the general correlation filter tracking algorithm is not ideal under many complex conditions, such as complex background, scale estimation, occlusion and so on. The problem of scale change is one of the most common problems in real scenes. This problem will seriously affect the accuracy of target tracking. When the target scale becomes smaller, redundant background i...

AI Technical Summary

Problems solved by technology

[0003] In order to solve the problems existing in the prior art, the present invention provides a real-time correlation filter tracking algorithm adaptive to scale changes, aiming at problems such as different scale changes and deformations, by initializing three filters with different ratios, and following In the frame, the original size of the target is used to predict the target position through the method in the original DCF. After the target position is determined, the three proportions of the target are sampled at three different scales, and the nine sampling targets and the three correlation filters are respectively Correlation filtering is performed on the corresponding scale, and the one with the largest confidence map is taken as the target scale of the frame to improve the tracking effect

Images