The Infrared Image Segmentation Algorithm Based On Cosine Theorem

Image segmentation is one of the most basic and important research contents in the field of computer vision.It provides a solid foundation for image analysis,image understanding and description.In recent years,a large number of scholars have found that the effect of the application of fuzzy set theory in the field of image segmentation is much better than the traditional method.However,there are still a lot of problems in the classical fuzzy image segmentation methods.In addition,the classic threshold segmentation algorithm can't correctly split the target class and the background class if there are large difference between them,and it will cause problems about over-segmentation or under-segmentation.According to the shortcomings of classical algorithm,we study the threshold segmentation algorithm based on fuzzy theory and improve it in this paper.The main research work is as follows:Firstly,we explored the application of fuzzy set theory in image segmentation,and summarized the general steps of image processing using fuzzy set theory: the first is to make images fuzzy,;then modify the fuzzy membership degree of elements in the fuzzy set;the last is to blur images.Secondly,on the basis of analyzing the deficiency of the classical threshold algorithm,this paper used fuzzy theory as the basic framework and combines the information entropy and the one-dimensional histogram information of the image to design a new membership function.The segmentation effect of the algorithm is greatly improved.Again,according to the classic threshold segmentation algorithm can't effectively segment image with the histogram showed a single peak feature,this paper designed a new membership function,and cosine theorem was introduced as the measure of the fuzzy distance for the first time,and a new threshold segmentation algorithm was presented.The algorithm in dealing with the histogram showed a single peak characteristic is better.Finally,the algorithms were realized by programming,and the results are verified by experiments.The experimental results were compared with the existing threshold segmentation algorithm,and also analysis the experimental results.