Research On Multiple Objects Detecting And Tracking Algorithm

Recently, video object tracking technology has been applied in many fields. Many scholars place their attention on it. Video object tracking have been used in Content Retrieving, Intelligence Surveillance, Perceptual Interface, Motion Analysis and so on. As an important part of the motion analysis system, the research in people detecting and tracking can provide potential uses in the future and benefit a lot to the society. Features-based tracking is frequently used in video object tracking; people also often recognize objects in daily life according to objects' features. If only we can get enough objects' features, then the objects can be tracked correctly.Occlusion among moving multi-object is a complicated problem in the video tracking research. Focused on it, the main work of this thesis is as following:1,Object Detection: we get the original image sequences which include moving objects by the camera. In this paper, we have reviewed three classic detection techniques: frame difference,optical flow technique and background subtraction, and have analyzed the advantages and disadvantages of them. We discuss a detection algorithm based on a background renewing method which based on the pixel's dynamic character in details. In the algorithm, we build the background modeling based on the consideration of the pixels whose values do not change sharply in a period of time as the background pixels. The foreground image which is expressed as a binary image is acquired by the difference between the currently image and the background model. We analyze the characters of this algorithm by experiments and discuss how to determine the values of parameters.2,Foreground Image Pretreatment: Morphological algorithms such as filtering and open operation are applied to the binary objects images in order to eliminate the disturb of the noises. An effective shadow detecting and eliminating algorithm is used to decrease the influence of the shadow of moving objects caused by the illumination. Discuss the connecting-region analysis algorithm in details and use it to locate moving objects in the scene.3,Object Tracking: Five states of moving objects in the scene are identified. Mark the related objects as one united target. Associated with the region and color-feature matching based algorithm raised in this thesis, the occlusion problem is solved effectively.The innovation works in the thesis are as following:1,A new tracking algorithm combining region corresponding-based algorithmand feature-based algorithm is presented to solve the problem. The occlusion was divided into two events, merging and splitting. Region corresponding was used to judge whether the merging and splitting events happened.2,To deal with the corresponding problem caused by splitting, further movingobjects' distinguishing is carried out with the application of a new color-feature matching to compute the similarity. The experiment results show that the proposed algorithm is efficient in computation and effective in resolving the multi-object occlusion.