基于分水岭算法的MELK图像分割  

Segmentation of MELK images based on watershed algorithm

作　　者：蔡念[1] 唐孝艳[1] 许少睿[1] 李防震[2,3] 

机构地区：[1]广东工业大学信息工程学院,广州510006 [2]中国科学院-德国马普学会计算生物学伙伴研究所,上海200031 [3]山东经济学院计算机科学与技术系,济南250014

出　　处：《计算机应用研究》2009年第8期3175-3176,3191,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(30600121);广东省自然科学基金资助项目(07301038)

摘　　要：提出一种基于分水岭算法的MELK图像分割方法,以解决信息量较小的MELK图像分割问题。综合应用多种技术以避免分水岭算法的过分割问题。首先应用高斯低通滤波器对MELK图像进行增强;采用加权差分算子求取梯度图像;应用Ostu方法进行对梯度阈值化,并进行分水岭变换;最后,应用RAG区域合并方法完成MELK图像分割。实验结果表明,该分割方法能够有效地分割信息量较小的MELK图像。This paper proposed a segmentation method for MELK images based on a watershed algorithm, which was applied to segment the MELK images that include little information. Employed several techniques to solve the over-segmentation problem in the watershed algorithm. Firstly, used a Gaussian lowpass filter to enhance the MELK images. Provided weighted difference operators to obtain the gradient images, and processed the gradient images by a watershed algorithm after thresholded them by the Ostu method. At last, segmented MELK images based on a region merging algorithm called RAG method. The experimental results indicate that the proposed method is feasible to segment the MELK images which include little information.

关 键 词：拓扑蛋白质组学 图像分割 分水岭算法 多抗原配体图谱图像 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]