Unmanned aerial vehicle target tracking method based on local image data learning

Abstract

The invention discloses a target tracking method based on local image data learning. The method comprises steps: dense sampling around the target is carried out, the obtained ones through sampling arehidden targets, and the ones beyond a sampling frame are discarded; based on local data, local features of the image are extracted, and the local data are learnt; a structural support vector machineis used as a classifier, and a discriminant tracking method is adopted; the saliency features of the image are calculated to acquire the position of the object; the tracking position of the structuralsupport vector machine and the object position obtained through saliency are fused to determine the final object position; and according to prediction, positive and negative support vectors are updated. The method disclosed in the invention can achieve good tracking effects in various scenes such as shielding, line changes and scale changes.

Description

technical field [0001] The invention relates to a method for implementing target tracking by using computer pattern recognition technology, in particular to a method for tracking an unmanned aerial vehicle target based on local image data learning. Background technique [0002] In traditional video target tracking, classic tracking algorithms such as subspace learning, sparse representation, multi-task learning, multi-instance learning, etc. It is to use various features to model the target, and then classify the potential target by comparing with the target or through the learned classifier to achieve the purpose of target tracking. They have achieved moderate success on common object tracking databases. According to the tracking method, tracking methods can usually be divided into generative and discriminative tracking methods. [0003] The generative tracking method is to model the potential target, and then compare it with the existing target by some measure, and the c...

AI Technical Summary

Problems solved by technology

[0008] Another target tracking method based on local image data learning, the steps include: (1) intensive sampling around the target, the sampled ones are potential targets, and those beyond the sampling frame will be discarded; (2) based on local data, Extract the local features of the image, and learn the local data at the same time, improve the stability of the target change, and improve the robustness to image noise and geometric deformation; (3) use the structural support vector machine as a classifier, and adopt discriminative tracking The method of improving the extracted local features has a good classification ability, and at the same time overcomes the problem of tracking loss caused by poor feature selection in the previous structure-based support vector machine classifier; (4) Calculate the salient features of the image to Obtain the position of the object; (5) through the tracking position of the structural support vector machine and the object position obtained by the salience, the two positions are fused to determine the final position of the object

Images