基于形态重构的豆粒分割算法

Image segmentation of soybean grain based on morphological reconstruction

在实现大豆图像豆粒的分离时,针对分水岭算法中容易出现过分割等问题,综合利用H-Dome形态学重构以及距离变换改善梯度图局部多个极小值问题,再使用分水岭算法找到粘结边界点,从而将颗粒分离。实验结果验证了此方法的有效性。

Distance transform and H-Dome morphological reconstruction were integrated to solve the problem of local multi-minimum in distance transform. Watershed algorithm was employed to find the border of adhered grains and to separate the adhered grains. The experimental results show that the integrated separation method is effective in image segmentation of soybean grains.