彩色图像分割方法综述

A Survey of Color Image Segmentation Methods

图像分割是图像分析中一个非常重要的预处理步骤,分割效果将直接影响到后续任务的有效性。彩色图像相较于灰度图像更接近人类的视觉特性,因此对彩色图像的研究更为重要。对当前比较常用的一些彩色图像分割方法进行了综述,阐述了基于阈值、基于聚类、基于区域以及基于特定理论的几类分割方法各自的优缺点和应用场景。最后根据基于过完备字典的稀疏表示能够刻画图像细节信息、实现图像最优逼近的特点,提出将其推广至彩色图像分割的研究思路。

Image is an important preprocessing step in image analysis,of whose effect directly affects the effectiveness of subsequent tasks.Color images are more closely related to the human visual characteristics than gray-scale images,so it bears more importance for study.In this paper,some segmentation methods of color image are reviewed,the advantages,disadvantages and application of segmentation methods based on clustering,region and class are analyzed.The paper then introduces the theory of sparse representation based on the over-complete dictionary,which can describe the detail information of the image and realize the optimal approximation of the image.The theory is proposed to be employed into color image segmentation.