基于广义图像与矢量量化的图像分割方法  

An Image Segmentation Method Based on the Generalized Image and Vector Quantization

作　　者：文超 李弼程[1] 平西建[1] 孙宏伟 

机构地区：[1]信息工程大学信息技术学院,郑州450002

出　　处：《信息工程大学学报》2000年第1期34-37,共4页Journal of Information Engineering University

摘　　要：图像分割的基本原则与矢量量化的思想是一致的 ,因此 ,可以利用矢量量化的技术对图像进行分割。考虑到图像象素点的空间相关性 ,本文将原图像与它的平滑图像组合起来 ,形成一个二元组 ,称之为“广义图像”。采用矢量量化技术得到广义图像的局部最优分割 ,从而实现对原图的分割。实验结果表明 :该算法具有很强的噪声抑制能力 ,取得了令人满意的分割效果。The principle of image segmentation is consistent with that of vector quantization.So the technique of vector quantization is used in image segmentation.For an image to be segmented,considering the correlativities among neighboring pixels,the paper combines the image with its smooth one,and generates a binary set that is called a “generalized image”.Locally optimal segmentation of the generalized image is derivedty using the technique of vector quantization,therefore the original image is segmented.Finally,experimental results and comparisons are given,and the new method proves to be effective and feasible.

关 键 词：图像分割 Otsu门限法 矢量量化 LBG算法 分裂法 

分 类 号：TN911.73[电子电信—通信与信息系统]