| Summary: | 碩士 === 嶺東科技大學 === 企業管理系碩士班 === 106 === Nowadays, visual detection system have had several cases of application, but in practical application and related documents, the visual detection system always use single means, which is easy influenced by light source and camera lens moving.
In this thesis, the system which we design for general video camera integrates three means. It is a set of being given first place to ZNCC and Edge, Color being assistant. Which make visual change detection become a more robust system. The experiment testified the system can not only judge the change of image, but also can distinguish the reason of the change of image. We have preliminarily improved the problem which are generated from using of the single change detection means. Improving common change detection system which generated some errors. Making the correct judgement to the change of image which generated by invasion of suspicious subject.
The case divided into 4 steps. First, choosing suitable background image. Second, divide background image and input image into a few blocks and using ZNCC to process. Third, add edge orientation in the results of ZNCC in order to deal with fortify accuracy of judgement. Fourth, make a further judgement according to color. Improving light domain’s results, improving the change detection system robust.
|
|---|
