摘要
针对分水岭算法存在的过分割问题以及VCH-F1切片图像的特点,提出了一种能够有效消除局部极小值和噪声干扰的分割方法。首先比较并选取彩色分量图像梯度信息的最大值,达到提取图像有效边缘信息的目的;然后提出基于多阈值分割的方法消除无效梯度信息;最后介绍了算法的步骤及结果。实验结果证明,基于该方法处理梯度图像进行分水岭算法处理可以得到准确的分割结果。
With regard to the over-segmentation of watershed algorithm and the characteristics of the slice images of the digitized Chinese No.1 female dataset,a new segmentation algorithm that can effectively eliminate local minima and noise disturber was presented.Firstly,effectively edge information was picked-up by comparing and choosing maxima from colour weight gradient images.Then,a method based on multi-threshold segmentation was proposed,which can eliminate inefficacious gradient information. Finally,procedures and results of the algorithm were introduced.Experimental results demonstrate that segmentation results are got exactly by watershed algorithm based on this menthod proposed.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第35期99-101,145,共4页
Computer Engineering and Applications
基金
国家自然科学基金( the National Natural Science Foundation of China under Grant No.30370393) 。
关键词
分水岭算法
彩色图像分割
多阈值分割
数字虚拟人
watershed algorithm
colour image segmentation
multi-threshold segmentation
virtual human
