Moving target tracking method based on template matching and deep classification network

Abstract

The invention discloses a moving target tracking method based on template matching and a deep classification network, and mainly solves the problems of low target detection speed and inaccurate tracking when a target is deformed in appearance and shielded in the prior art. The implementation scheme comprises the following steps: 1) establishing a dual-residual deep classification network, and training the dual-residual deep classification network; 2) extracting a template network and a detection network from the dual-residual deep classification network; 3) extracting template characteristicsby using a template network; 4) extracting detection characteristics by using a detection network; 5) carrying out template matching on the template features on the detection features to obtain a template matching graph; 6) determining a target position according to the template matching graph; (7) updating template characteristics according to the target position, (8) judging whether the currentframe is the last frame or not, if yes, ending target tracking, otherwise, taking the updated template characteristics as the template characteristics of the next frame, and returning to the step (4).The method has the advantages of high tracking speed and high accuracy, and is used for tracking the violent deformation and illumination change video target.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a moving target tracking method, which can be used for tracking video targets of such types as severe deformation, camera shake, scale change, and illumination change. Background technique [0002] The main task of moving target tracking requires to learn a tracker when only the initial frame information of the target to be tracked is known, so that the tracker can accurately predict the position of the target to be tracked in the next frame of the video sequence. With the continuous deepening of people's understanding of the field of computer vision, moving object tracking has been widely used and developed in this field. Due to the continuous application of deep learning in the field of image classification and image segmentation, deep learning methods have gradually been applied to the field of object tracking. Compared with the manual feature extraction method ...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the weighted result of the target template and the background template is used as the confidence value, which fails to reflect the fluctuation of the response of the target to be tracked, and the recognition ability of the trained classifier is not strong enough. , Long-term and accurate tracking cannot be achieved when the target moves quickly

The disadvantage of this method is that it is necessary to determine the sample type of each local area of ​​the image by setting a threshold. When the target to be tracked is occluded to a large extent, the target sample or background sample will be misclassified, resulting in an updated model. Cannot continue to track target accurately

Images