| The object detecting and tracking technology of video sequences is an important research topic in the field of computer vision. Its core technology is the means to extract the location information of the target in the video sequence and get the target trajectory. At present, Objects Detecting and Tracking algorithm was mature in the static background. In recent years, with the mobile robot and intelligent equipment development, the application of the mobile camera increased gradually, then the background is no longer stationary, commonly used in static background video target detection and tracking algorithm is no longer applicable. So it is very important to study the target detection and tracking algorithm in dynamic scenes.Thesis analyzed the research status of object detecting and tracking technology in the video sequence at this stage. This article studied and discussed the moving target detection based on compensation method of background and the target tracking based on the algorithm of TLD. For some problems and deficiencies, I completed the following work:Firstly, Thesis learned and researched the compensation method of background. The paper presents an improved compensation algorithm based on the SURF algorithm to make the estimation of global motion parameter more exact. The algorithm used SURF feature to match so that it can get a more accurate match data. And it used RANSAC algorithm and the concept of residual filtered the error matching points.Finally, it used morphological filtering and the concept of connected domain to filter the noise. In order to ensure real-time matching,the paper adopts a more rapid and efficient FAST detecting algorithm when using SURF algorithm to detect feature points. Secondly, the paper learnedand researched TLD algorithm. In order to improve the the real-time of algorithm, the detection module of TLD algorithm is added to the color classifier. Color classifier is a classifier based on the color information that is presented by this article. It is placed in foremost position of hierarchical classification to reduce the pressure of variance classification.Finally, this paper combined TLD algorithm and background compensation algorithm to achieve the automatic tracking of moving objects.Thesis uses Visual Studio 2008 to realize the improved algorithm in the windows environment. A variety of video sequences are tested under dynamic background. In the moving target detection,experimental results show that the improved algorithm can improve the compensation accuracy and has a good detection effect in dynamic background. In the aspect of moving object tracking,the improved TLD algorithm can achieve the target tracking in dynamic background,and the processing rate is doubled.The algorithm has good real-time performance and application value. |
