摘要
在大米籽粒检测中经常会有大米籽粒重叠和粘连的现象。将粘连在一起的大米籽粒群分离是图像处理中一项重要而困难的问题。对这一情况,提出了一种有效的分割算法,先对图像进行二值化处理,然后对图像进行数学形态学变换,通过腐蚀运算将大米籽粒群图像层层剥离,直至粘连籽粒分开,再使用膨胀运算和八链码方法恢复出原始的各个大米籽粒个体,从而将大米籽粒群分离开。实验证明,该算法能成功地分离开粘连大米籽粒。
In detection of the rice grain,there is often a phenomenon of rice grain overlapping and adhesion.The separation of adhesion grain image is an important and difficult problem.According to this problem,in the paper,an efficient segmentation algorithm was proposed,e.g.,the binary image processing,and then transformation of the image morphology,stripping the rice grain base image layer by erosion operation until the grains were separated from adhesion,finally restoration of the original individual grains of rice by dilation and eight direction code,by which the rice grain could be separated from the group.Experimental results showed that the algorithm could successfully separate the rice grain from adhesion.
出处
《农业网络信息》
2010年第10期18-21,共4页
Agriculture Network Information
基金
河南工业大学科研基金(编号07XJC015)
关键词
粘连颗粒
图像分割
数学形态学
边缘提取
particle adhesion
image segmentation
mathematical morphology
edge extraction
