Research On Repair Algorithm Of Damaged Textile Images And Three-Dimensional Reconstruction

In the era of rapid development of computer technology,image restoration technology,as an important branch of digital image processing,and it has a wide range of applications in daily life.Cultural relics are part of the civilized society.After the damaged cultural relics are restored,the historical features of the time can be reproduced.The restoration of cultural relics plays an important role and significance for the inheritance of cultural roots.As an important part of cultural relics,textiles have already appeared in ancient China.However,due to the characteristics of its own material,which is easy to rot and difficult to preserve.Therefore,it has important historical significance and practical application value to restore the image of ancient damaged textiles.Image restoration technique includes two methods based on variational partial differential equations and texture synthesis.In this paper,Criminisi image restoration algorithm based on texture synthesis is selected to repair damaged textile images.In order to improve the quality of textile image restoration and enhance the application of cultural relic restoration in 3D reconstruction,Criminisi algorithm was further improved and 3D reconstruction of textile image was carried out.The main tasks as follows:(1)Since there is no method in the existing image restoration technology that can clearly distinguish different types of images,this paper proposes a method based on image gradient operator and difference,which accurately distinguishes texture and non-textured images by objective data.(2)In the original Criminisi algorithm,confidence items and data items were used to calculate the priority of the repaired blocks.However,the repaired image by this calculation method has a structural break and a texture blur.In this paper,according to the strong structure of textile image,a new variable is created using the edge strength and angular intensity of the structure tensor,which is used to enhance the structural of the image during the repair process.The original priority calculation adopts the multiplication method,which makes the value of priority quickly become zero and affects the repair effect.Therefore,in this paper,data items,confidence items and new variables are added to each item of the calculated priority to adjust the repair effect of the image.(3)In the original Criminisi algorithm,a fixed-size repair block was used to repair the image with matching errors and edge discontinuities.According to the periodicity of the textile image,an adaptive repair block size selection method is proposed.In this paper,a new image quality evaluation method is created based on peak signal-to-noise ratio and structural similarity.The size of the repair block is changed during the repair,and the new quality evaluation method is used to find the best repair block.(4)In this paper,the restored textile image is three-dimensional reconstructed using modeling software to make the image more stereoscopic and authentic.The experimental results show that the proposed algorithm can solve the problems of matching confusion,structural discontinuity and texture blur in the image,and highlights the significance of textile image restoration in cultural relics restoration.There are 41 figures,7 tables and 65 references in this article.