Research On Image Restoration Algorithm Based On Texture Synthesis

With the rapid development of computer science and technology,the application fields involved in image processing are becoming more and more extensive.Image restoration is an important branch in image processing,and it is also a research hotspot in the field of image processing in recent years.In fact,image restoration is a process in which the known information in the image is used to fill the damaged part according t o some rules and get the result consistent with human visual perception.The application of image restoration involves all aspects of people's life,such as photo restoration,removal of unwanted objects,and so on.In the daily life,it is inevitable that large areas of the image will be damaged.Therefore,the large area damage image restoration algorithm has important research value and significance.In this paper,the large-area damaged natural image is used as the repair object.The improvement of the Criminisi algorithm is improved,and an image restoration algorithm based on contour constraint is proposed.Although the Criminisi algorithm can achieve better repairing effects in repairing large-area broken images,when repairing damaged images with texture and structure coexisting,there will still be structural line breaks and texture extension caused by matching errors.The main reason lies in the shortcomings of the Criminisi algorithm itself.For example,in the later stage of the repair process,due to the irrationality of the priority calculation formula,the value of priority will gradually become unreliable,and the order of repair will become unreliable;the matching criterion only refers to the color similarity of pixel blocks which leads to a matching error with sigle message.In view of the two shortcomings listed above,this paper improves the priority calculation formula and matching criterion respectively,firstly changing the algorithm in the priority calculation formula.Secondly,the structural information in the structural similarity measure is introduced into the matching criterion to improve the matching accuracy of the structure.Although the above improvements have achieved better results compared with Criminisi algorithm,when repairing damaged images with missing important structures,the human eye can still perceive that there is a break at the edge of the repair,which is not smooth enough.In this regard,further improvements are proposed,and the image segmentation technology is used to extract the structural information of the image,and the damaged structure is preferentially repaired.Then the reconstructed edge constrains the priority of image restoration,so that the restored image achieves a natural continuous effect.Finally,the simulation results are evaluated in real with the two methods of subjective and objective evaluation.The results show that the improved algorithm can effectively maintain the consistency of the repaired results and the rationality after texture filling,as well as in line with human visual perception.