Three-Dimensional Model Construction And The Design And Implementation Of Dynamic Target Recognition Algorithm

With deeply study of computer vision, VLSI development and improvement of modern industrial and military automation, various engineering needs based on target detection and tracking are emerging, such as intelligent video surveillance, vehicle tracking and so on, which promotes greatly the domestic and international academic research. Target detection and tracking technology has been applied in military and civil fields, and also has been become research focus in automatic control, computer vision and pattern recognition. At present, the traditional target detection algorithms can’t effectively detect the targets in the dynamic background, and tracking algorithms are used under the conditions of the environment and lack of the universal. Therefore the research and design of target detection under the dynamic background and tracking algorithm are the main objective of this article.The main contents are three-dimensional modeling of large outdoor scenes, the design of the scene targets detection and tracking algorithms and the implementation of the two algorithms. In the three-dimensional modeling part, it uses the three-dimensional modeling software3ds max to create the scene, sets the scene’material and texture, designs the animation, renders the scene and finally saves the.avi video file. In the part of target detection, firstly it pretreatments the video file, secondly it researches the target detection algorithms under static background and the global motion estimation and background compensation techniques. It introduces the block matching algorithm and uses it to change the dynamic background into static background and designs the three difference method to overcome the shortcomings of the two difference algorithm. Next, it uses the three difference method and the block matching algorithm to finish the target detection under the dynamic background. It chooses the CamShift algorithm as the tracking algorithm and designs the experiments for the different environments. Finally, it defines the target detection and tracking models and inserts the two models into OpenCV video surveillance framework.