微分运算与形态学深度图像分割比较

首先研究了深度图像基于微分不变量和形态学边缘检测算子的两种常用分割方法，接着提出了一种改进的形态学水线区域分割方法，并用同一物体的深度图像进行对比分析实验。实验结果表明：改进算法的速度较基于微分不变量的分割方法快，但略慢于基于形态学边缘检测算子的分割方法；改进算法提取的边缘较前两种分割方法光滑，分割结果与人的主观视觉感知具有良好的一致性。

Compare of range image segmentation methods with differential coefficient operation and binary morphology

Firstly the two range image segmentation methods in common use with differential coefficient operation and morphology edge-detecting operator are researched.Then an improved segmentation method based on morphology waterline region is put forward,and the experiments to compare and analyze the segmentation result and computing speed of the three methods for range image of the same object are finished.The experiment results indicate：The speed of the improved method is faster than the method with Differential operator,while it is a little slower than the segmentation method based on morphology edge-detecting operator;the edges extracted by the improved arithmetic are smoother than the edges extracted by the two traditional segmentation methods,and its segmentation result is well consistent with subjective vision of human.