Video target real-time tracking method and system based on depth feature fusion and adaptive correlation filtering

Abstract

The invention discloses a video target real-time tracking method and system based on depth feature fusion and adaptive correlation filtering. The system comprises a feature extraction module, a feature fusion module and a correlation filtering module. The feature extraction module uses a lightweight network model to extract multi-layer depth features, and the real-time performance of a target tracking task is ensured. The feature fusion module provides a multi-layer depth feature fusion strategy of typical correlation analysis for the problem that independently extracted multi-layer features are not complete enough in representation of a target. According to the method, the target expression capability and the target and background distinguishing capability are improved, the feature redundancy is reduced, and the calculation amount of subsequent related filters is reduced. The correlation filtering module provides a correlation filter updating strategy based on response value dispersion analysis for solving the tracker drifting problem caused by challenges that the target is deformed, shielded, moved out of the view, and is rotated existing in a target tracking task, filter template updating is carried out in a self-adaptive mode, and the specific problems are relieved.

Description

technical field [0001] The invention relates to the technical field of target tracking, in particular to a method and system for real-time tracking of video targets based on deep feature fusion and adaptive correlation filtering. Background technique [0002] The video target tracking task initializes the target position in the first frame, and then uses the tracking algorithm to analyze and process the video frame frame by frame to predict the target position. In recent years, methods based on deep learning and correlation filtering have attracted extensive attention in the field of video object tracking, which has promoted the improvement of object tracking performance. [0003] CF2 [1] Independently extract multi-layer depth features from the deeper VGG-16, and send them to the subsequent correlation filter for prediction calculation, creating a precedent for the combination of deep learning and correlation filtering. However, due to the use of more complex feature extr...

AI Technical Summary

Problems solved by technology

However, the multi-layer deep features extracted from the lightweight model have insufficient ability to represent the target, limited ability to distinguish the background and the target, and there is redundancy between features, resulting in additional calculations.

At the same time, target tracking tasks often face specific problems such as target deformation, target occlusion, target moving out of view, and target rotation. Using existing correlation filtering algorithms is prone to tracker drift or even tracking failure.

Images