The Digital Image Processing Based On Fuzzy Theory

When the image 1S generated,transmitted and stored,it often exposed tO all kinds of noise interference,and this would serious impact on the visual effects．Therefore,many kinds of image processing technologies have been proposed,and the imagefiltering is one of the key technologies．Image filtering technology must reduce thenoise,and at the same time,as much as possible tO maintain．the image details,andget preparation for the follow-up processing．Through the study of the traditional image filtering algorithm,analyses itscharacteristics and shortcomings,we have proposed a simple algorithm based on thefuzzy weighted hybrid filtering,which focused on dealing with the uncertainty of thenoise,as well as the defects of the traditional algorithm．This algorithm based onfuzzy logic and then calculates the final results through these data．In this paper．we first use membership functions and fuzzy rules tO detect pixelstO determine the noise type．and use different filtering methods according todifferent types of pixel noise,and then we create a number of dimension and theimages the same size matrix to sign the relevant noise types．denoted by the matrixsymbol called the noise matrix．First．According to the noise matrix of the signs．Weuse adaptive median filter algorithm to the impulse noise filter．and then use obliquecross—neighbor fuzz3'weighted mean filter tO the Gaussian noise for the overall filter．We select the neighborhood of 8 pixels of oblique cross 5×5 filtering windowcenter of pixels to extend the symmetric pixel．In the selected eight pairs of symmetricpixels and the center of each pixel on the symmetry"point of difference between theminimum grayvalue of a pixel as the pixel of neighbor pixels to achieve betterprotection of image in detail．At the same time．We use the characteristics and imageinformation to adj ust the pixel values of the fuzz?,weighted mean filter algorithm tooptimize the membership function to achieve better noise reduction capability'．The series of comparing experiments have shown that this algorithm reducesnoise and at the same time as much as possible tO maintain the details of the image．The algorithm has fully considered the relevance of the pixels and the uncertainty ofthe noise...