Research On The Method Of Mural Segmentation Based On Graph Cuts Theory

As an extremely precious artistic heritage in Chinese national culture,murals not only have the characteristics of early origin and wide spread,but also are amazing art classics after thousands of years of accumulation and precipitation.However,the existing murals generally have fuzzy edges,unclear Outlines,overall fading,obvious noise and other problems due to direct contact with air.The traditional protection method is an irreversible process,and the digital protection method has attracted more and more attention.Mural segmentation is an important process for the digital protection of murals,and the result of the segmentation directly affects the later restoration work of the murals.The use of intelligent information processing technology can be a good way to achieve automatic segmentation of murals.Combining the characteristics of existing mural images and digital image processing technology,this paper proposes two models of mural segmentation based on Graph cuts Theory,which lays a foundation for mural restoration in the later period.Aiming at the long-term existence of existing murals and the phenomenon of lack of damage,using a single algorithm for processing is prone to under-segmentation and over-segmentation problems.Based on these problems,This paper proposes a new model of mural segmentation combining k-means clustering and Grab Cut algorithm.The initial clustering of k-means algorithm was completed by using the peak information of gray histogram after image smoothing,and then the pre-segmented image was used as the initialization information of Grab Cut algorithm to achieve the final segmentation of murals.Experiments show that this method has good performance in mural segmentation.In order to solve the problem that the overall contour is difficult to locate in the segmentation process due to the blurring of some murals and the unclear edges,this paper also proposes a mural segmentation algorithm with enhanced edges.The algorithm uses adaptive wavelet denoising to remove the noise in the mural image and fuses the mural contours extracted by Sobel operator and Canny operator to enhance the edges,and finally achieves the segmentation of the mural.Experimental results show that this method can not only effectively eliminate the noise interference,but also get a good segmentation effect for murals with unclear edges.Based on the above research,this paper uses Matlab R2017 b to design and implement a mural segmentation prototype system based on the graph cut theory,and intuitively shows the segmentation effect by calling the segmentation algorithm.This paper provides two segmentation algorithms based on the characteristics of existing murals,which have strong applicability in the segmentation of ancient murals and provide a new idea for the segmentation of murals.The prototype system realizes the visualization of mural segmentation and provides a practical value for the digital protection of murals.