期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2015
卷号：3
期号：6
DOI：10.15680/ijircce.2015.0306101
出版社：S&S Publications
摘要：Image inpainting is a method of restoring lost, deteriorated, damaged or selected portion of an image bymaking use of the neighboring information. Inpainting should be such that, the inpainted image when viewed by anycasual observer should feel a visually pleasing continuous flow of data in and around the inpainted region. This paperproposes a method for image inpainting using DEMA (Diffused Expectation Maximization Algorithm) and GBS(Graph Based Segmentation) and compares it with the inpainting using GBS only. Here an Exemplar-based in-paintingmethod is adopted, which iteratively search the neighboring region (source region) and fill the missing region (targetregion), with the most suited patches from the source region. The proposed method uses graph based segmentationalong with diffused expectation maximization algorithm to improve the performance of inpainting. In Graph BasedSegmentation method the spatial information in the source region is utilized .This method selects the parameter valuesof the robust priority function and thereby determines the most suited patch size and reduces the search region. Thenumber of segments obtained from GBS is used in DEMA thereby we get a DEMA segmented image which is onceagain graph based segmented in order to get the parameter value of the DEMA segmented image. This parameter valuealong with the segmented image is passed through an exemplar based inpainting algorithm forming a better inpaintedimage. Certain parameters to evaluate the performance of the proposed algorithm is also done. Experimental resultswith a number of test images shows the effectiveness of the proposed method.
关键词：Diffused Expectation Maximization; Graph Based Segmentation; Exemplar Based Inpainting;Inpainting
