Visual Tracking For Mobile Robots Based On Camshif Algorithm

Vision has the advantage of broad sensing area and full information. With the development of image processing technology and the ability of computer, the visual tracking technology is becoming a main way in robot domain. The vision-based tracking has maken great progress in the last two decades. Vision-based mobile robots identify and trace moving target through segmenting sequent images from visual sensor into moving object in the space domain and modeling, identifying and tracking the object in the time domain.With the support of project of National Scientific Fund and project of Excellent Scientist Fund in Hubei, the research on mobile robot visual tracking based on Camshift algorithm is put forward. Through studying image processing theory, especially the visual tracking theory and method, this paper designs a good robust visual robtot tracking system in which the Camshif algorithm is employed and improved, and the same time proposes a simple and effective mothod of explaining the images into robot-controling instructions.Through the study of this subject, some results and conclusions are drawn:(1) This paper analyzes the features of color and shap in detail. It also compares the methods of extraction these features. Finally the color feature is chosen as the feature for visual tracking.(2) This paper does some research on visual tracking based on existing tracking algorithms. It puts forward a good robust tracking algorithm based Camshift algorithm which is fit to tracking common objects. And some results have been obtained at the aspect of the study of enhancing the accuracy of visual tracking by employing Kalman filter and processing backproject image.(3)This paper discusses the possibily of producing robot-controling instructions by the means of visual range measurement with the aid of the laser, and proposes a simple method of producing robot-controling instructions.(4) Under the Visual C++ development environment, usingOpenCV libs, a real-time effective visual tracking system is implemented. And some system-testing experiments are carried out.