基于自适应射线群的图像边缘检测算法

Image Edge Detection Algorithm Based on Adaptive Ray Group

对于存在大量噪声、目标边界模糊且粘连的浮选泡沫类图像,分水岭及阈值法难以准确分割。为此,提出自适应射线群算法检测泡沫边缘,仅访问图像一次,即实现种子区域的提取。去噪后,从种子区域的几何中心位置对称发射出多条射线,根据射线的灰度分布曲线自适应提取泡沫的边缘,并修正边缘。实验结果表明该算法可解决分水岭算法的过分割及不准确分割等问题。

Owing to the problems of some images, such as bubble image, in which a lot of noises exist and objects mutually adhere and have high similarity. It is difficult to detect edge by watershed and threshold method. Ray-based image segmentation method is proposed, seed areas of image are exlracted by visiting image only one time. After filtering noises, a number of symmetric rays from geometric center of seed regions are launched, the edge of bubbles is gotten by gray value of curve graph of every ray. Experimental results show that this method can amend fuzzy edge, and solve over-segmentation and poor accuracy problem.