A method for visual tracking through spatio-temporal context

Abstract

The present invention provides a method for visual tracking through spatio-temporal context, comprising the following steps: Step 1: Initialize parameters; Step 2: Train context-aware filter to obtain position model; Step 3: Train maximum scale response value of scale correlation filter to obtain Scale model; step 4: classifier output response map; discriminant correlation filter generates peak sidelobe ratio corresponding to peak of response map; step 5: compare response map peak to peak sidelobe ratio, if response map peak is greater than peak sidelobe ratio , then introduce an online random fern classifier for re-detection; if the peak value of the response graph is smaller than the peak sidelobe ratio, update the position model of step 2 and the scale model of step 3; if the peak value of the response graph is equal to the peak sidelobe ratio, continue to maintain the current Visual tracking state; step 6: apply the updated position model and scale model to the next frame tracking; return to step 4.

Description

technical field [0001] The present invention relates to the visual tracking field of computer vision, in particular to a method for visual tracking through spatiotemporal context. Background technique [0002] Visual tracking is an important research hotspot in the field of computer vision, and it is widely used in video surveillance, autonomous driving, car navigation, and human-computer interaction. The purpose of tracking is to accurately estimate the position of subsequent frames when the position of the first frame is known. Although it has made great progress in recent years, it still faces challenges from many external factors. For example, in the long-term tracking process, the target usually experiences some external disturbances, such as occlusion, illumination change, deformation, scale change and out-of-view, which will affect the accuracy of visual tracking. [0003] Tracking tasks are generally divided into location estimation and scale estimation, which are ...

AI Technical Summary

Problems solved by technology

[0004] Although the existing techniques achieve the desired tracking results and perform well for long-term tracking, when the target object undergoes complex appearance changes (such as severe occlusion) and disappears in the current frame, it will introduce some wrong background information and will be Passed to the next frame, long-term accumulation will deteriorate the quality of the tracking model and eventually cause tracking target drift

Images