| Summary: | 碩士 === 中華大學 === 電機工程學系碩士班 === 102 === With the increasing demands of functions of Digital TV, there are more and more factors, which have essential influence upon product quality and operating stability, so that quality testing and insurance become more and more important. Therefore, how to effectively reduce the timing cost of product testing is an issue of great importance. In this study, we shall derive an automatic intelligent algorithm to replace labor test. For an MPEG frame, based on the theory of image processing, we shall first obtain edge information of each macro block, which is the first feature of the considered macro block, by using the gradient information of the picture. Meanwhile, to obtain the second feature of a macro block, we calculate the differences between the direct current term of the discrete cosine transform of the target macro block and those of its neighboring blocks. Based on the feature vector consisting of these two features, each macro block can be categorized into different clusters by applying the fuzzy C-mean algorithm so that we can effectively sort out the mosaic macro blocks.
|
|---|
