단순선형회귀분석과 에지 검출기에 근거한 영상 잡음의 분산 추정

초록

When using various devices for capturing images, it is a general phenomenon that is the images contamination by several reasons. So it is too difficult to prevent its images from adding noises. If we process some images, we should detect the noises and get rid of that or reduce. Measuring relative size of noise, this study estimates a level of variances in the noise which are contaminating images. At first, use the Sobel edge detector for selection of non-edged block, and apply a testing method in regressor of simple linear regression for confirming value of pixel in the blocks. At last estimating the noises by mean of standard deviation of block. The simulation result has been confirmed that improvement in precision of the proposed estimator was compared with robust median estimator. And the suggested estimator was also verified that was not so influenced as Gaussian noise and quantization noise. In addition, this study confirmed the efficiency performance of removing noises, using sigma filter by estimated deviation of estimated noises.

키워드