Research On Exemplar-based Digital Image Inpainting Algorithms

Digital image inpainting is the technique whose purpose is to restorea partially damaged or missing image to resume its integrity and originalvisual effect. It is of great importance in a variety of applications, suchas protection of cultural relic and paintings, restoration of damagedimages caused by network transmission, removal of caption, scratches andtarget objects in images and videos etc.This paper summarizes the fundamental and research status of image inpaintingtechnique. Researching the performance of exemplar-based image inpaintingalgorithm,analyzing and summarizing the usual improved approaches.On that basis,two inpainting algorithms are proposed:(1) The proposed image inpainting algorithm base on region segmentation andinformation of example. Through the whole image segmentation get dividedregions,Provide effective local structural features and similarity of the texture regionsegmentation.By effective structureal features, from the known structure of the curve tothe unknown region extends to obtain the missing region structural information with acurve contour.In the image restoration process through different structural feature to getdifferent segentation region,and choose a reasonable similarity matching blocks to fill.Experimental results show that the algorithm is effective. Visual effect than otheralgorithms have significantly improved.(2) The proposed texture synthesis algorithm based on exemplar mosaic. Thisalgorithm is by looking for relevant degree between two images of sample blocks toachieve the highest image mosaic.Small scale images can be spliced in differentdirections for a large area of the image,finding paths of minimum error tooverlapping portion. In the process boundary seams add human visual associated rights values to pixels, by finding the correlation between the two images to achieve thehighest sample block image mosaic. The texture features are given the small sampleblock, creating large area of similarity and surface texture of continuous,thus reducingtime and improving the quality of image mosaic.Experimental results show that thealgorithm is effective and retain the original mural art style.