基于改进分水岭算法的粘连颗粒图像分割

Image Segmentation of Adhering Particle Based on Improved Watershed Algorithm

对颗粒物质实验中粘连颗粒的分割,传统分水岭算法容易产生过分割现象。为解决该问题,设计了一种有效的改进分水岭算法。该算法先采用数学形态学重建滤波平滑图像噪声及内部小细节,然后使用Otsu方法对图像进行阈值分割,并对得到的二值图像作欧氏距离变换,将像素点位置信息转换为灰度信息,最后对距离图利用分水岭算法得到最终分割图像。实验结果表明,该算法获得了较满意的分割效果,解决了目标粘连现象对后续分析、测量产生干扰的问题。

During the segmentation of adhering particle in the experiments, traditional watershed algorithm has oversegmentation prob lem. To solve the problem,an effective and improved watershed algorithm was proposed. Firsdy ,it used morphological reconstruction fil tering to smooth the image, and then it used Otsu to do the image threshold segmentation and calculated the binary image＇ s Euclidean dis tance. Finally, the final segmentation image was obtained by means of the watershed image segmentation. The results in the experiments show that the algorithm gets satisfactory segmentation effect and it is successful to solve the interference problems of adhering disk.