Study Of Image Segmentation Algorithm Based On Fuzzy Logic

Image segmentation as image analysis, recognition and understanding prerequisites is characterized into disjoint regions by different features. There are various kinds of image segmentation methods, the thesis combines fuzzy theory with thresholding segmentation or edge detection to improve the quality of image segmentation.First, the thesis analyzes the threshold method based on ambiguity, it has been improved some of the shortcomings of the algorithm such as poor adaptive and bandwidth that needs someone to determine. We have solved these two problems by histogram transformation and fuzzy rate curve. Experiments show that the effect of image segmentation can better restore image details. Then, the thesis introduces a two-dimensional image fuzzy entropy combined gray information and spatial information of image through analyzing the characteristics of one-dimensional image fuzzy entropy. And it has proposed a two-dimensional fuzzy entropy weighted thresholding segmentation by improving optimization algorithm method of exhaustion and combining with intelligent optimization algorithm. Experiments show that the algorithm segments image accurately and has some noise immunity.Finally, the thesis introduces some classic operators of the image edge detection method, analysis the classic fuzzy edge detection method for enhancing by Pal and King presenting. For some problems existing in transform algorithm, it makes the corresponding tries and improvements. It introduces the Edge Confidence and returns the modified parameters, edge detection results show good results.