Study Of Tracking Algorithms In Moving Target A Complex Background

The study of target tracking algorithms in moving target under a complexbackground is actually a way to locate and track a target on the image sequence. Inthis paper, we introduce the development and present situation of image targettracking technology. At the same time, we briefly describe the classification of theexisting common target tracking methods, and introduce the application of imagematching and bionic technology in target tracking technology. Aiming at somelimitations of current target tracking algorithm in practical application, we put theemphasis on two kinds of target tracking algorithms which based on transform domainand neural network.The registration precision of existing target tracking algorithm will decline rapidlywhen the registration images have a large scale deformation. To solve the problem oftarget’s size and rotation changes under a complex background, we combine Log-polartransform and Phase Correlation Algorithm. This algorithm can effectively perceivethe multiple scale and rotation angle of a target; we analyze the algorithm performancein different kinds of influences. And we use the favorable center registration ability ofWeight Average-removed Normalized Cross Correlation to improve this algorithm,that makes the proposed algorithm effectively ameliorate the center sensitive problem.The algorithm has great performance of small amount of calculation, rotation andscale invariance and anti-noise. Simulation results show that this method can improvethe robustness of the tracking algorithm.Because of their own limitations, traditional pattern recognition methods bringmany restrictions to their actual applications. This paper introduces the idea ofhigh-dimensional bionic neural network and designs the multi-weightshigh-dimensional bionic neural network. According to the different weights ofnetwork, we verify and compare their effects and application characteristics. Based onthe above study, we select a appropriate overlay method to the application of imagetracking algorithm. Combine with the improvement plan of anti-shelter, we realize thetracking process of image sequence with a better adaptability, accuracy and efficiency.