基于分水岭算法的绿色作物与背景分割研究  

Study on the Segmentation of Green Crops from Background Based on Morphological Watershed Algorithm

作　　者：饶洪辉[1] 姬长英[2] 

机构地区：[1]江西农业大学工学院,江西南昌330045 [2]南京农业大学工学院,江苏南京210031

出　　处：《安徽农业科学》2009年第29期14483-14484,共2页Journal of Anhui Agricultural Sciences

基　　金：教育部重点科研项目(03091);江西省教育厅项目(GJJ-08177)

摘　　要：分水岭算法作为彩色图像分割手段的一种方法,具有运算简单,性能优良,能较好提取运动对象轮廓和准确得到运动物体边缘等优点。应用分水岭算法研究了绿色作物及其背景的分割,首先通过数码相机拍摄的一幅640×480田间青菜真彩色图像,在matlab中采用分水岭分割算法处理图像后提取其绿色分量,再用数学形态学闭运算处理后可以较好地分割绿色作物与背景。针对结果中存在的过分割现象,采用先计算图像的形态学梯度,再用分水岭算法分割可以使结果得到有效改善。As a algorithm of true color image segmentation ,watershed segmentation has less computation, better performance, which can also obtain profile and edge of mobile object . The algorithm was used for segmentation green corps from background for its performance. At first, a true-color greengrocery image of 640 pixels width and 480 pixels height obtained by digital camera, which was processed by morphological watershed in the software of madab. The greengrocery and background was segmented by extracting its green vector, then processing by morphological closing. Direct application of the watershed segmentation algorithm to process image generally led to the over-segmentation phenomenon. The results could be effectively improved by computing morphological gradient of the image then using morphological watershed.

关 键 词：分水岭算法 绿色作物 图像分割 

分 类 号：TP242.64[自动化与计算机技术—检测技术与自动化装置]