摘要
材料样品的粒度分析是材料结构分析中一项基础而又重要的内容。在分析颗粒图像特点和不良分割原因的基础上,通过构造基于势能比较的评价函数,提出一种改进的分水岭算法来分割颗粒,以避免传统分水岭算法和势能阈值算法分割不当的问题。实验结果表明,该改进算法得到的分割结果正确性、鲁棒性更好,因而其粒度分析可靠性更高。
The granulometry of material samples is a basic and important work in structural assessment of materials. Based on analyzing the features of grain images, this paper proposes an improved watershed algorithm with an energy-comparison-driven function. The new approach can avoid bad segmentations which occur in the conventional watershed and the energy-thresholding watershed. Experimental results demonstrate that the proposed algorithm is more precise and robust and that the granulometric statistics thus are more reliable.
出处
《电子器件》
CAS
2008年第4期1369-1372,共4页
Chinese Journal of Electron Devices
关键词
粒度分析
分水岭算法
距离变换
势能比较
granulometry
watershed algorithm
distance transform
energy comparison
