generally received financial compensation of the cufflinks

Common edge detection operator to check the rate of change of the neighborhood of each pixel and gray level can be quantified.
Sobel operator sub-[21] [Ba by two convolution kernels, as shown in Figure 3.4. Each point in the image of these two nuclear convolution, a check is usually the vertical edge response maximum, while the other horizontal edge response. The maximum of two convolution as the output bit. The operation result is an edge gradient image. Figure 3.4 of Sobel operator to the pearl cufflinks texture information, it is also manifested by a series of continuous gray scale. Thus, using the Sobel operator to operator, get the edge information of pearl cufflinks and the surface texture of pearl cufflinks. With inspection report by the authority, the consumers then negotiate with the seller, generally received financial compensation of the cufflinks.
After sobel operator computing the results obtained in Figure 3.4. shows. Binarization can see that the surface texture of the pearl cufflinks on a gradient map to get better performance, while most of the useless information to be filtered. It is understood that the city Quality and Technical Supervision Inspection Institute has the ability to test the cufflinks. At this time, pearl cufflinks the global OSTU binarization, you can get Figure 3.4. By the effect of the figure can be found in the gradient map of the global maximum between-class variance binarization, a good addition to the edge of the outside noise are filtered out, but also most of the texture information is lost. Analyze the reasons for the way the gradient, the edge of the pearl cufflinks and a white background is the biggest difference between the two types of texture information with a white background difference is not large, so the gradient map.  The two features most useful for this change is the rate of change and direction of the gray, they are respectively the gradient vector and the direction of.