摘要
针对于矿物浮选过程泡沫图像目标气泡粘连、大小形状灰度不均及目标相似度高的问题,分水岭及阈值法难以分割此类图像,笔者提出一种射线分割方法。首先进行目标定位,从种子区域的几何中心位置对称发射出多条射线,并获得各条射线中像素的灰度值变化曲线图,根据各条射线的灰度变化曲线图的波谷位置得到泡沫的边缘。此方法能够去除噪声泡沫,并参照一个泡沫的清晰边缘部位对其模糊边缘进行修正,使其形成真正符合泡沫形状的光滑边界。
Owing to the problems in bubble image that the size and shape are uneven,there exists the phenomenon of bubble mutual adhesion and the objects have high similarity,the watershed and threshold methods are hard to be used for the segmentation of bubble image,so a ray-based image segmentation method is proposed.First,the object is located.Next a number of rays are emitted symmetrically from the geometric center of seed region and the grey value change curves for every ray are obtained.Finally according to the trough location of grey value change curve,the bubble edge can be gotten.This method can remove noise bubble,and can correct fuzzy edge of one bubble according to its clear edge part to form a real smooth boundary in line with the bubble shape.
出处
《北京石油化工学院学报》
2010年第1期36-40,共5页
Journal of Beijing Institute of Petrochemical Technology
关键词
射线分割
曲线图
种子区域
光滑边界
ray segmentation curve graph seed region smooth boundary
