Estimating the Image Segmentation Number via the Entropy Gap Statistic

To estimate the images segmentation number, a model was proposed based on the entropy gap statistic. The ldquogap statisticrdquo method had been advanced by Tibshirani R. etc. firstly. The ldquogap statisticrdquo method compares the change in within cluster dispersion with that expected under a uniform null distribution. In the paper, the entropy gap statistic mainly considers the change of entropy in a set of data. In the images segmentation method based on the entropy gap statistic, the element of the set of data is the gray value of an image, the gray distribution of reference image is a uniform distribution, and it analyses the characteristics of the image segmentation model via the entropy gap statistic, compared with the images segmentation method ldquogap statisticrdquo method.