Target detection in stationary background

Abstract

The application of object tracking and detection are important parts in many computer vision systems, the reliable classify for an image sequence as foreground or background is still a hot research project in nowadays. In this experiment, we are trying to find the moving object by using webcam, and three basic image processing algorithms are used to get the best result. We found the results obtained by using these three basic algorithms were not perfect, it depends a lot on the environment setting, such as illumination condition. After combined the background subtraction and optical flow algorithms, we found the result was much better compare with these three individual methods. The basic principle of target detection is to compare the current image with a reference image which is call the background image, from the comparison result to define whether there is a moving target in the image by checking the changing in the image, pixel by pixel. The details of the operation for each method have been presented in this thesis.