计算机技术与发展Issue(2)：12-16,21,6.DOI:10.3969/j.issn.1673-629X.2016.02.003
基于卡通提取的自然图像分割
Natural Image Segmentation Based on Cartoon Component Extracting
摘要
Abstract
The traditional edge based level set image segmentation may be over- or under segmentation for nature scene,rich in texture. To suppress the texture on the influence of the image segmentation,propose a new model for natural image segmentation,which follows results of R. O. F. edge-preserving model and Y. Meyer texture-preserving model,using it to extract images to cartoon component and texture component. According to the pixel variation characteristics of texture in the spatial domain,express texture as the divergence of function gradient and build a new cartoon-extracting model in the image domain. In addition,design diverse equation for the model by fixed point iteration algorithm,and the convergence condition by the relationship between the contour of the object and the edge of car-toon component. The experimental results show that the cartoon component extraction algorithm inherits the advantages of traditional total variation algorithm,implementing the approximated constant expression for the texture region,retaining the object contour,bluring weak edges inside the object,to a certain extent,suppressing the effect of texture on segmentation algorithm.关键词
图像分割/卡通分量/纹理分量/保边/水平集Key words
image segmentation/cartoon component/texture component/edge preserving/level set分类
信息技术与安全科学引用本文复制引用
郭娟,何坤,周激流..基于卡通提取的自然图像分割[J].计算机技术与发展,2016,(2):12-16,21,6.基金项目
四川省科技支撑项目(2013SF0157) （2013SF0157）
