Small low-flight target visual detection tracking system and method thereof

Abstract

The invention discloses a small low-flight target visual detection tracking system and a method thereof. The system comprises a video data input unit, a video preprocessing unit, a training data construction unit, a detection model training unit, a target comparison screening unit, a detection correction unit, a reference frame initialization unit, a sample library dynamic construction unit, an online learning unit, a position refinement unit, a decision control unit and a tracking result output unit. The method comprises the following steps: constructing a target detection network, and comparing and screening targets; carrying out target tracking online learning; and dynamically constructing a classifier training sample library, and performing target tracking position refinement. The system and the method have the following advantages: the tracking drift condition of the tracking target caused by factors such as shielding, scale change and illumination can be effectively relieved, androbust target tracking can be realized; the capability of updating reference frame features in time according to target changes is achieved; and meanwhile, error tracking caused by updating the reference frame features can be avoided by introducing a feature point matching algorithm.

Description

technical field [0001] The invention relates to the technical field of flying target tracking, in particular to a small low-flying target detection and tracking system based on neural network and online learning and a detection and tracking method thereof. Background technique [0002] At present, there are already some related methods of joint detection and tracking and some tracking methods suitable for slow and small targets. The existing method realizes the short-term target tracking through the correlation filter method, and uses the target detection method based on the neural network to realize the function of relocating the target when the tracking fails. Related patents and research technologies are as follows: [0003] Chinese invention patent, titled: A robust long-term tracking method based on correlation filtering and target detection, application number: CN201910306616.4; the invention realizes target tracking through correlation filtering method, and uses one-...

AI Technical Summary

Problems solved by technology

However, this method does not make further processing after the target tracking failure, and the accuracy needs to be further improved

[0005] Combining the above existing technologies, none of them can solve the problem of extreme imbalance of the target background in single target detection and tracking, the network performance has not yet reached the optimal level, and there is still room for further improvement in the tracking accuracy of small targets

Images