binarization principle value of, also known as gray partitions

Thresholding method, depending on the local information of the image or the overall information, can be divided into context-free u), this method in 1978, the calculation is simple, stable and effective, has been widely used. Pattern recognition perspective, the optimal threshold should result in the target class and the class separation performance of Beijing.
 This performance class variance to characterize, for the introduction of the variance within stare Chu class variance stare; and overall variance stare; and define the criteria for the three equivalent measurements: Diao = poison in practical use, often using the following simplified formula: Formula seven four in view of computational considerations, it is generally by optimizing the third criterion to obtain the threshold. Thus, the awareness of safeguarding the cufflinks should be increased.  This method also has its defects, the experiments revealed: failure of the very hour when the ratio of the size of the target and background image. (d) = Zi Mountain stare (a ‖) 2 + (% ‖) 2 formula (3.11) which, fixed 2 for the largest variance between the two categories, Class A probability, that the Class A average gray % Class B the probability that the mean gray of Class B, ‖ for the image overall average gray. Draw out the cufflinks with flexibility
Therefore, first the image binary of the image into the background and foreground thread analysis is useful prospect information. binarization principle value of, also known as gray partitions (Threshold), the general image of the gray scale divided into only two gray values, setting a gray value Any image gray level is greater than it would make it 255 gray values below the set value, they make it as dark spots, so can get a binary image. However, the proof testing of the cufflinks are often provided with the sales by businesses.