Stochastic Fuzzy Discrimination Information Measure Cost Function in Image Processing

A new cost function based on stochastic fuzzy discrimination information measure is introduced in this paper. Focusing on their significant parts, this cost function is used to find the optimal value of threshold for denoising image. It is, in fact, an extension of fuzzy entropy cost function by the present author. Multivariable normal distribution is used for creating focus on significant parts of an image. At the end, the results of this cost function are compared to previous ones by applying it to some images. By using multivariate normal distribution on the images as the cost function weight, the center of the image is more considered by the algorithm. Consequently, the best results will be produced by the new cost function.