摘要
在医学图像的目标识别和三维显示过程中,边缘检测是十分重要的一环,检测的准确性将直接影响到后续的治疗.虽然现在已有许多边缘提取算法,但这些方法用于医学图像分割时很少能同时满足速度快、最优化以及所检测边缘连接甚至封闭等要求.本文提出一种基于动态规划(DgnamicProgramming,简记为DP)的边缘检测的快速算法,在算法中采用局域窗技术加快搜索速度,由分段搜索自适应地实现快速最优化巨连续的目标边缘检测.算法在模型图像及医学图像中实施,取得了满意的效果.
In this paper, a 2D dynamic programming based model is proposed for edge detection, while local-window technique is used with the optimal parameters determined adaptively in the window. The whole algorithm is sped up by contour tracing segment by segment, and finally gets optimal and connected contours. Experiments on simulated images and medical images show that the proposed approach can achieve a fast and ideal edge detection operation.
出处
《计算机学报》
EI
CSCD
北大核心
1997年第12期1063-1068,共6页
Chinese Journal of Computers
关键词
边缘检测
图像分割
医学图像
图像处理
Edge detection
dynamic programming
object recognition
image segmentation.