作　　者： ; ;

机构地区： 浙江警察学院

出　　处： 《计算机应用与软件》 2010年第7期256-258,共3页

摘　　要： 针对目标与背景颜色接近,目标区域粘连度较高难以分割的情况,提出了一种基于阈值处理和分水岭分割相结合的改进的圆形目标图像分割算法。提及的算法分成两个步骤:第一步采用了3次阈值处理相结合的改进算法,较好地完成了目标和背景颜色接近情况下的目标提取;第二步采用了改进的分水岭分割算法,改善了分水岭分割之前的距离变换和求种子点过程,较好地实现了粘连度较高情况下的目标区域分割。实验结果表明该方法对圆形粘连目标的分割有较好的效果。 In some images,the colours of the object and of the background are very close that it is difficult to separate those conglutinated object regions.According to this situation,this paper introduces an improved image segmentation algorithm for circular object based on combined threshold selection and Watershed Segmentation algorithms.Here we divide the algorithm into two steps：First,we use an improved target selection algorithm in which the threshold selection method is applied three times,and the object＇s picking up is well accomplished in condition of its colour closing to the background＇s;Second,we use an improved Watershed Segmentation algorithm,which meliorates the progress of distance transformation and determining the seeds before using Watershed Segmentation algorithm,and has well realised object regions segmentation in condition of higher conglutination.Experimental result shows,the method has good effect on segmenting conglutinated object,especially the circular one.

关 键 词： 阈值 图像分割 分水岭

领　　域： [自动化与计算机技术] [自动化与计算机技术]