Object Tracking Method Fusing Convolutional Network Features and Discriminant Correlation Filters

Abstract

The invention discloses a target tracking method integrating convolutional network features and discriminative correlation filters. An end-to-end lightweight network architecture is built to train convolutional features by learning rich flow information in consecutive frames, improving feature representation and tracking accuracy. The correlation filter tracking component is constructed as a special level in the network to track a single image block. During the tracking process, the target block and multiple background blocks are simultaneously tracked, and the target and its surrounding environment are identified by perceiving the structural relationship between the target and the surrounding background blocks. A model is established for the high-degree part, and the target tracking effect is measured by the peak sidelobe ratio and the peak value of the confidence map. In the case of large-area occlusion, extreme deformation of the target shape, and dramatic changes in illumination, the background part of the discrimination is automatically used. to locate.

Description

technical field [0001] The present invention relates to a target tracking method integrating convolutional network features and discriminative correlation filters. Background technique [0002] Object tracking is a fundamental problem in computer vision. A common process for this problem involves inputting a continuous video image, initializing the object of interest with a bounding box in the first frame, and estimating the target in subsequent frames. The location of the object. Visual tracking is an important technology in computer vision, which has a wide range of applications in security protection, only monitoring, human-computer interaction, and automatic control systems. [0003] In recent years, many researchers have conducted extensive research on discriminative correlation filter (DCF)-based visual object tracking and have made great progress. With the development of the method, the existing algorithms can well solve the motion tracking problem in the simple mot...

AI Technical Summary

Problems solved by technology

Since the derivation work is carried out in the Fourier frequency domain, the present invention not only retains the characteristics of high CF efficiency, but also uses the convolution feature to improve the feature representation method of the target. The limitation of the weak structure further solves the problem of convolution features in the case of large-scale occlusion. Underperforming issues, significantly improved tracking accuracy and rate

Images