Current Issue Cover
模糊相关图割的非监督层次化彩色图像分割

尹诗白,孔垂涵,王一斌(西南财经大学;四川师范大学)

摘 要
目的 常用的最大模糊相关分割方法存在划分数需预先确定,阈值的分割结果存在孤立噪声,无法对彩色图像实施分割的问题。方法 为此,提出了基于模糊相关图割的非监督层次化分割策略来解决该问题。算法首先将图像划分为若干超像素,以提高层次化图像分割的效率;随后将快速模糊相关算法与图割结合,构成模糊相关图割2-划分算子,在确保分割效率的基础上,解决单一阈值分割存在孤立噪声的问题;最后设计了自顶向下层次化分割策略,利用构建的2-划分算子选择合适的区域及通道,迭代地对超像素实施层次化分割,直到算法收敛,划分数自动确定。结果对Berkeley分割数据库上300幅图像进行了测试,结果表明算法能有效分割彩色图像,分割精度仅次于半监督分割方法,明显优于非监督分割方法,且运行时间至少提高了40%。结论 本文算法为最大模糊相关算法在非监督彩色图像分割领域的应用提供了指导依据。
关键词
Unsupervised hierarchical color image segmentation through fuzzy correlation and graph cut

Yin Shibai,Kong Chuihan,Wang Yibin(Department of Economic Information Engineering,Southwestern University of Finance and Economics,Chengdu;Department of Engineering,Sichuan Normal University,Chengdu)

Abstract
Objective The frequently-used maximum fuzzy correlation method has some limitations, i.e. partition number needs to be preset, the results have isolated noise and maximum fuzzy correlation approach cannot be extended to color image segmentation. Method To address these problems, an unsupervised hierarchical color image segmentation through maximum fuzzy correlation and graph cut is proposed. First of all, we over segment the color image into superpixels for improving the efficiency of hierarchical image segmentation. Then, we combine the fast fuzzy correlation with graph cut to form a bi-level segmentation operator which can suppress the isolated noise caused by single threshold-based approach. Finally, a top-down hierarchical segmentation approaches has been designed. By iteratively performing this bi-level segmentation operator on selected color channel and regions, the superpixels can be segmented in a hierarchical manner and partition number is achieved automatically. Result The performance of algorithm has been tested on the 300 color images in Berkeley segmentation database, and the results demonstrate the presented scheme ranks only second to semi-supervised segmentation schemes and is superior to the existing unsupervised segmentation methods. Besides that, the running time is 40% shorter than that of compared methods. Conclusion Our method provides an important reference for the application of the maximum fuzzy correlation algorithm in the field of unsupervised color image segmentation.
Keywords
QQ在线


订阅号|日报