A Target Tracking Method Based on Unmanned Vehicle Application

Abstract

A target tracking method based on the application of unmanned boats, comprising the steps of: acquiring the image sequence of the target, using the preset filter f(z)=W within the set range where the position of the target in the previous frame is the search center T Z, the maximum response position is searched, and the maximum response position is used as the candidate position center of the target in the current frame; several candidate frames are obtained by using multi-scale sliding window sampling around the candidate position center, and all candidate frames are calculated using a structured SVM classifier Score, the candidate frame with the highest score is used as the prediction result of the target in the current frame; judge whether the target is occluded; if the target is occluded, online learning will not be performed, and the preset filter parameters and structured SVM classifier parameters will not be updated; if the target is not is occluded, update the preset filter parameters; update the structured SVM classifier parameters; repeat steps S1‑S5 until the last frame of the image sequence. The invention can track stably for a long time.

Description

technical field [0001] The invention relates to the technical field of unmanned boat applications, in particular to a long-term stable tracking method for targets based on unmanned boat applications. Background technique [0002] An unmanned boat is an unmanned surface ship that can be used in various environments, especially in environments that are not suitable for manned ships and in relatively dangerous environments. The demand for boats is gradually increasing. In the autonomous navigation of the unmanned boat, the stable tracking of the target is the technical basis for the automatic obstacle avoidance of the unmanned boat. At present, there are mainly two methods of target tracking: [0003] (1) Target tracking method based on key point matching [0004] The target tracking method based on key point matching usually extracts key points with invariant features in the target area, regards the target template as a collection of some key points, extracts key points in ...

AI Technical Summary

Problems solved by technology

[0007] The success of the method based on key point matching is closely related to the detection method of key points. Too many key points will affect the efficiency of algorithm execution, and too few key points will affect the accuracy of the algorithm. Moreover, when the appearance of the target changes drastically, or the background It is more complicated, and when it is not very different from the target, this algorithm will lose the target, resulting in tracking failure

Although the tracking method based on target detection can solve the problems of deformation and rotation of the target well, the complexity of the algorithm is high, and it is difficult to achieve real-time, which limits the practical application range of such methods.

[0008] Although there are many target tracking methods, in the autonomous navigation of unmanned boats, there are many types of targets to be tracked (such as cruise ships, sailboats, floating objects, reefs, etc.), which may be large or small, and often With the scale change of the target and the occlusion between the targets, the current target tracking method cannot be well adapted to the real natural scene under the premise of ensuring real-time performance.

[0009] In summary, although there are many methods related to target tracking, it is difficult to apply them to the automatic obstacle avoidance of unmanned vehicles due to the robustness and real-time performance of the algorithm.

Images