一种基于TV正则的图像卡通纹理分解方法  

Image Cartoon Texture Decomposition Based on TV Regularity

作　　者：李军 何瑞强[1] LI Jun;HE Rui -qiang(Xinzhou Teachers University,Xinzhou 034000, Chin)

机构地区：[1]忻州师范学院

出　　处：《忻州师范学院学报》2018年第2期4-6,30,共4页Journal of Xinzhou Teachers University

摘　　要：文章研究将一幅图像分解为卡通部分和纹理部分。图像的卡通部分通常由几何的、平滑变化的成分组成,而纹理部分所包含的是图像的细节和噪声。给出TV-L1变分模型,它由L^1数据忠诚项和总变差(TV)正则项组成;将该模型用于u+v的图像卡通纹理分解;最后,实验说明TV-L1模型应用于图像卡通纹理分解的效果。Paper mainly studies the decomposition of an image into a cartoon part and a texture part. The cartoon part of an image is u- sually composed of geometric, smoothly -varying component,while the texture part contains the details of the image and noises. Present a TV - L1 variation model ,which consists of L＇ data fidelity item and Total Variation （TV） regularization term. Then the model is used to obtain u ＋ vimage cartoon texture decomposition. Finally, the experimental results show that the model is applied to the cartoon texture decomposition of image.

关 键 词：卡通纹理分解 总变差 不动点 

分 类 号：O187.1[理学—数学]