水泥窑生料球图象分割方法的研究

Image Segmentation of Globular Marerial

以水泥窑生料球为背景，分析了球形物料图象在灰度空间的特征，对一些经典算法进行了颗粒分割实验．理沦分析和实验结果表明，利用局部特征值进行总体象素归属判决将使颗粒图象的边界不连续，或边界描述失真．为此提出一种由局部特征值直接提取球形物料边界的图象分割方法．

The feature of the globular on the grey space and greadient space is analyzed in thispaper, the segmentation test of particle image on a few classical algorithms were made. Thetheoretic analysis and the experiment result show that using local feature value make the detectsthe edge of the particle directly according to local eigenvalue is proposed. The experimentresult shows that the image segmented by this method have good edge continuity, and trueegde. the method is suitable for the segmentation of the cumulate big particle's image.