Image segmentation algorithm based on improved information entropy and grey relational degree analysis

The thresholding algorithm based on maximum entropy is an important method for image segmentations. The grey relational degree analysis indicates the correlative degree exactly between two factors. To improve the shortage of the original thresholding method of maximal entropy, some new methods are proposed in this paper. First, parameterize maximal entropy segmentation principle, and evaluate the segmentation effect based on Gray-level Contrast and grey relational degree analysis to select parameters. Secondly, introduce the exponential form of entropy and weight it, which reflects gray distribution and select the parameters of weight based on Gray-level Contrast and grey relational degree analysis. Finally, give a deformation of maximum entropy based on high frequency grayscale, fully consider the effects on segmentation of high frequency grayscale. The experiment results indicate that the thresholding value, which is defined by these improved methods in this paper, can obtain superior segmentation results.