摘要：为提取颗粒图像的边缘,详细讨论非极大值抑制图像(NMS)灰度统计值的变化特点,提出一种基于颗粒图像的Canny算子双阈值计算方法。通过实验发现NMS图像中"谷"两端的灰度值为最佳双阈值。结果表明:该方法比已有的自适应算法更加准确,能够最大程度地去除噪声,保留有效边界。在图像法测量颗粒形态的过程中,该方法相对基于最大熵原理或Otsu方法的Canny算子更具有可靠性和稳定性。

关键词： 颗粒测量；Canny算子；图像法

Abstract： In order to get the clear boundary of particles， the changes of gray level for non-maxima suppression image （NMS） were discussed in detail and a new method on calculating the dual-threshold for Canny operator based on particle images was proposed. The gray values at both ends of valley function curve in NMS were found the perfect dual-threshold. Compared to Canny operator based on maximum entropy or Otsu， this method had much more reliability and stability in the application of particle measurement using image. 

Keywords： particle measurement； Canny operator； imaging