形态学梯度重建的改进快速分水岭算法

Developed Watershed segmentation Based on Morphological Gradient Reconstruction

提出一种基于形态学梯度重建的改进快速分水岭分割算法。该方法在形态学梯度图像基础上,利用形态学开闭重建运算对梯度图像进行重建;定义了分水岭的强度指标--落差,基于此对快速分水岭算法进行了改进,并将此算法结合图像重建进行了仿真实验,整个分割过程无需进行分割后的区域合并处理,降低了分割的复杂性。仿真实验证明,改进的算法无论从消除过分割还是区域轮廓定位等性能方面,均具有较好的分割效果。

A method of watershed segmentation based on morphological gradient reconstructing was proposed. First,morphological gradient image was obtained. Second, opening and closing by reconstruction operators were employed to reconstruct gradient image. Then, a guide line of watershed＂ s strength named ＂fall＂ was defined, based on which, watershed segmentation arithmetic was developed, then we applied the developed method to the reconstructed gradient image. The whole segmentation process needed no post-segmentation which in some degree reduced the complexity of the segmentation. Results of simulation showed that the developed method can efficiently eliminate over-segmentation, and hold the position of region contours without evident bias.