블록 방법에 의한 적응적 영상 잡음제거 알고리즘

초록

Despite the development of video equipments, it is almost impossible to completely control the noise caused by various reasons. Such noise may deteriorate the image quality and adversely affect image utilization such as image recognition or analysis. Usually we approach this problem as a method of removing the noise based on the sample variance. However, it is not easy to distinguish between noise and feature with sample variance only, and introduces the concept of orientation of feature. When the sample variance is calculated in various orientations, the sample variance in the direction of the feature is significantly smaller. In this paper, various orientations are considered as a block method using two pixel windows. And we confirmed the homogeneity of variances by using the ratio of maximum variation and minimum variation in each direction. And we proposed a feature statistic that reflects the degree of feature, reflecting strongly the degree of feature when the noise is weak, and weakly reflecting the degree of feature when the noise is strong. As a result of the simulation, it is confirmed that the overall image quality is improved by adaptively removing the noise by the algorithm proposed in this study.

키워드