基于边缘检测双边滤波的彩色图像去噪.pdfVIP

8 V l .38 N . 8 2010 8 ACTA ELECTRONICA SINICA Aug. 2010 张 闯, 迟健男, 张朝晖, 王志良 ( , 100083) : , . ( CNN) , CNN , CNN . , . , , , , , , , . . , . : ; ; ; ; ; ( CNN) : TP391 : A : 0372-2112 ( 2010) 08- 1776-08 Removing Noise of Color Images Based on Edge Detection and Bilateral Filter ZHANG Chuang, CHI Jian-nan, ZHANG Zha-hui, WANG Zhi-liang ( School of I nf ormation Engineeri ng , University of Sci ence and Technol ogy Beij ing , Beji i ng 100083, Chi na ) Abstract: F r the disadvantages f c l r bilateral n ise filtering, this paper presents an integrated appr ach t address these issues based n edge detecti n and bilateral filter. First, a new c l r edge detecti n alg rithm f adaptive partiti n is derived fr m cellular neural netw rk ( CNN) m del. This alg rithm inherits the advantage f accurate p siti ning f CNN gray edge detecti n, and can deal directly with c l r images. Next, an edge filter is pr p sed f r image enhancement, which satisfies different stages f den ising thr ugh tw types f edge detecti n. On this basis, rem ving n ise f c l r images uses f the impr ved bilateral filter. The sc pe f n ise is l calized by n n-ant-i n ise edge map t reduce the den ising pr cess and image blur . Average m de f bilat- eral filter is impr ved t reduce the weight f n ise and the impact f high-frequency n ise. Finally, c l r images are enhanced ac- c rding t the edge map after filtering. The experimental results sh w that the pr p sed appr ach can effectively reduce the n ise while preserving and enhancing edges. Key words: edge detecti n; bilateral filtering; c l r image; edge filter; den ising; cellular neu