B样条函数的图像分割分水岭算法研究  

Watershed algorithm research of image segmentation based on B-spline function

作　　者：余成波[1] 秦华峰[1] 

机构地区：[1]重庆工学院远程测试与控制技术研究所,重庆400050

出　　处：《计算机工程与设计》2008年第18期4777-4779,共3页Computer Engineering and Design

基　　金：重庆市自然科学基金计划重点项目(CSTC2007BA2023);重庆市教育委员会科学技术研究基金项目(KJ070620)

摘　　要：针对传统分水岭算法存在的过分割和对噪声敏感问题,提出了一种能够很好地抑制噪声、剔除图像的伪边缘、准确定位图像边缘信息的方法。采用高频强调滤波对梯度图像进行增强,利用B样条函数对增强后的图像进行多次拟合,最后对拟合的曲面进行分水岭分割。实验结果表明,通过该方法处理的梯度图像再进行分水岭变换,有效避免了过分割问题,同时准确定位了图像边缘信息,提高了分割精度。With regard to the over-segmentation of traditional watershed algorithm and the problems of sensitivity to noise, a new algorithm that can effectively restrain noise, eliminate image edges and detect the image edges exactly is presented. Firstly, morphological gradient image edge is enhanced by high frequency emphasize filter effectively. Then, the enhancement image is fit by B-spline function many times. Finally, watershed segmentation is used for the smoothing surface. Experimental results demonstrate that watershed segmentation with the proposed algorithm resolves the phenomena of over-segmentation well. The scheme can detect the image edges exactly and improve segmentation precision.

关 键 词：高频强调滤波 B样条函数 曲面拟合 分水岭算法 图像分割 

分 类 号：TP911.73[自动化与计算机技术]