基于边界剥离分水岭算法的重叠颗粒分离研究

针对采集的洗煤厂煤尘图像颗粒重叠严重的问题,运用传统图像二值化方法处理速度慢、不适用于低对比度指纹图像。分水岭分割算法对噪声敏感和易于产生过分割问题,为了提高图像分割的效率和鲁棒性,提出一种新的基于边界剥离距离变换分水岭算法的图像分割方法。该方法通过将图像分割为不相交的N个子窗口,并求出各个窗口对应像素的灰度均值,经改进动态阈值法进行二值化处理,得到的二值化图像经边界剥离距离变换获得灰度图像,最后利用分水岭算法对图像进行分割处理。数值实验结果表明,与传统分割方法比较,此方法处理效率高、鲁棒性好。不但成功地解决了分水岭存在的过分割问题,还大大提高了算法的搜索效率;可以快速准确地分割出目标,是一种有效的图像分割方法。

Boundary Stripping Watershed Algorithm Based Overlaping Particles Separation

Overlapping coal dust particles is a serious problem, we can use traditional binary image processing method , but its speed is slow; and watershed segmentation algorithm is sensitive to noise and prone to have over-segmentation problem. In order to improve image segmentation efficiency and robustness, a new image segmentation method is produced based on dynamic threshold watershed algorithm. Through this method, the image is divided into N disjoint sub-windows, and the each sub-window corresponding pixel gray value is obtained. The improved dynamic threshold binarization method and watershed algorithm are used to process image segmentation. Numerical results show that, Compared with the conventional segmentation method, the new method is high efficiency and the robustness is good. Not only It solve the over-segmentation problem successfully, but also improve the algorithm search efficiency greatly. The new method can segment the target quickly and accurately, so it is an effective method to resolve the overlapping dust particles image.