초록

용어보기논문에서 용어와 풀이말을 자동 추출한 결과로, 시범 서비스 중입니다. AI-Helper

Abstract ▼ AI-Helper

The exemplar-based image inpainting technique is to fill missing region in an image, and it is widely used for recovering the scratches, removing the textures and deleting unwanted objects. The exemplar-based inpainting algorithm is composed of two procedures : selecting a patch to be filled from th...

The exemplar-based image inpainting technique is to fill missing region in an image, and it is widely used for recovering the scratches, removing the textures and deleting unwanted objects. The exemplar-based inpainting algorithm is composed of two procedures : selecting a patch to be filled from the missing region, and inpainting the patch using image information in the known region. In the patch selection procedure, patch priority should be defined to encourage the filling-in of patches on the structure with higher priority. In the patch inpainting procedure, similarity between the selected patch and a patch in the known region should be defined to determined which patch is to filled-in among those in the known region. In the conventional exemplar-based inpainting algorithm, gradient and sum of absolute difference are used for the priority and the similarity measure, respectively. These measures are based on the brightness of the pixels in the image. This paper is to propose a method which uses the structural information rather than the brightness levels for both the priority and similarity measure in exemplar-based inpainting. This method helps retaining structural information, and thus the inpainted result looks more natural than the conventional way. Structure matrix has been used for obtaining the structural information. Experiments show that the proposed method produce more natural image than that of the conventional exemplar-based inpainting proposed by Criminisi.

Keyword