Research On Spectral Clustering Image Segmentation Algorithm Based On Interval Fuzzy Theory

With the continuous development and advancement of science and technology,computer vision and pattern recognition have received more and more attention in recent ten years.The theory of this subject is perfected and applied successfully in more and more fields.Clustering analysis is a key technology in the field of computer vision and pattern recognition.Because traditional clustering algorithm converges to the local optimal solution,and then introduces the spectral clustering algorithm,it has no restriction on the distirbution of the sample space and can converge the global optimal solution,and also can obtain good image segmentation results.Although spectral clustering algorithms have shown good performance in many fields,the spectral clustering algorithm will have a high spatial complexity when the data set is large,and there are still many problems in extracting eigenvectors and constructing similarity matrices problem.An interval fuzzy spectral clustering ensemble algorithm for color image segmentation(IFSCE)is presented in this thesis.Firstly,the color histogram is obtained by the just noticeable difference color threshold method.Then the interval fuzzy similarity measure based on color feature is constructed by utilizing the interval membership degree and the image are grouped by normalized cut criterion under the similarity matrix produced by interval fuzzy similarity measure.Finally,the segmentation results with different optimal fuzzy factors combination are integrated to get the final result.This thesis mainly aims at the following points:(1)Explain the research significance and purpose of selecting this subject and the work done in this dissertation.Subsequently,the introduction of interval fuzzy theory and spectral clustering algorithm,the description of the clustering process and solve the traditional clustering algorithm to converge the local optimal solution.The image segmentation and image model are described and understood,and the necessity of applying interval fuzzy clustering algorithm to image segmentation is described.(2)The construction of similarity matrix in spectral clustering algorithm is a key issue.In this thesis,by introducing the color histogram and the interval fuzzy theory of image,an interval fuzzy clustering image segmentation method based on color histogram is proposed.The method uses fuzzy theory to construct the measure of similarity,and then uses the similarity matrix produced by interval fuzzy similarity measure to obtain the final result by using the normalized cutting rule to segment the image.Due to the introduction of color histogram and fuzzy theory,the segmentation performance of traditional spectral clustering algorithm is improved.(3)Validity evaluation index can get more accurate segmentation results.The effectiveness index based on the interval fuzzy clustering is obtained by the clustering evaluation index,and the fuzzy factor combination is evaluated by the effectiveness index to obtain a group of optimal fuzzy factors.Each pair of the optimal fuzzy factor combination can be obtained by constructing a similarity matrix.Finally,different results were integrated by clustering integration algorithm.An algorithm of spectral clustering integrated image segmentation based on interval fuzzy theory was proposed.Compared with other traditional spectral clustering algorithms,this algorithm improved the segmentation results.It shows that the algorithm has greatly improved the computational complexity and the segmentation effect.