초록

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

Abstract ▼ AI-Helper

In this paper, we propose a new method to resolve the dependency problem of search sequence and the similarity calculation problem in the Criminisi's inpainting algorithm. Previous methods have two problems: they partially compare two patches and they give different results depending on the search s...

In this paper, we propose a new method to resolve the dependency problem of search sequence and the similarity calculation problem in the Criminisi's inpainting algorithm. Previous methods have two problems: they partially compare two patches and they give different results depending on the search sequence. To solve the problem, we propose a new method that restricts the search area and compares two patches by using the difference between the mean of RGB values in the patch. We improved the problem by using the difference of pixel mean values in the similarity computation. Also, we proposed a new equation for confidence term to solve the 'dropping effect' problem. In the experimental results from our four test data sets, we verified that our method gives more similarity degree between original images and result images even when the damaged area is wider and bigger. Our method shows the potential of the image inpainting in practical use to remove unnecessary objects in pictures that captured in daily life.

주제어