基于图象腐蚀和区域生长的矸石图象分割算法  

Segmentation Algorithm of Gangue Image Based on Image Eroding and Region Growing

作　　者：宋永宝[1] 孙伟[1] 赵峻[1] 权强强[1] 叶家良[1] 

机构地区：[1]中国矿业大学信电学院,江苏徐州221008

出　　处：《工矿自动化》2009年第12期29-32,共4页Journal Of Mine Automation

摘　　要：在实际的矸石分选过程中,要求的目标主要是将大块的矸石分选出来。但是,在拍到的胶带运行过程图片中,往往夹杂着很多小的煤颗粒或是矸石颗粒,如果不滤除掉这些小的颗粒,会影响矸石分选的效果和效率。针对上述问题,提出了一种融合图象腐蚀和区域生长的矸石图象分割算法。该算法首先对采得的原始图象进行图象压缩,然后对压缩的图象进行直方图均衡化,通过设定合适的腐蚀半径对原图象采取图象腐蚀处理,并选择合适的种子和阈值对腐蚀后的图象进行区域生长,最后将处理后的图象和原图象做"与"运算,得到边缘清晰的大块矸石图象。仿真结果表明,该算法能有效分割出大块矸石,且经图象腐蚀后的区域生长阈值的取值范围明显变大,对其它边缘模糊图象具有一定的参考作用。In actual process of gangue separation, the main target is separating big gangue from coal and gangue. But images captured from running belt usually adulterate a lot of small coal particles or gangue particles, which will influence effect and efficiency of gangue separation if no filtering these particles. For above problems, the paper proposed a segmentation algorithm of gangue image based on image eroding and region growing. The algorithm firstly compresses the original image, then takes the compressed image for histogram equalization, makes image eroding process by setting suitable radius of eroding, and selects suitable seeds and threshold to make region growing. At last, It takes the processed image to compute with the original image, which can get big gangue image with clear edge. The simulation result showed that the algorithm can separate large gangue effectively, and the range of threshold of region growing of eroded image is bigger, which has reference function for images with bad edge.

关 键 词：矸石 分选 图象腐蚀 区域生长 图象分割 阈值 

分 类 号：TD94[矿业工程—选矿]