基于计算机视觉的纸张填料粒径分析方法

针对现有纸张填料粒径分析过程中使用的激光衍射法存在的缺点,提出一种基于计算机视觉的纸张填料粒径分析方法。将获取到的扫描电子显微镜图像进行预处理,去除噪声等影响统计粒径分布的因素,对图像中存在的填料颗粒粘连比较严重的部分采用改进的分水岭算法进行图像分割,进而利用快速连通域标记及欧拉数算法确定填料颗粒的粒径分布。最后将采用计算机视觉方法计算的填料颗粒粒径分布结果与采用激光衍射法得到的结果进行对比。在放大1500倍沉淀碳酸钙填料颗粒图像上的实验结果表明,采用计算机视觉方法计算的纸张填料粒径分布与激光衍射法测得的结果基本一致,中位径误差小于1%,累计粒度分布数达到85%时,误差在6%左右。

A Particle Size Analysis Method for Paper Filler Based on Computer Vision

Aiming at the existing disadvantages of laser diffraction method used in particle size analysis of paper filler, a method for particle size analyzing of paper fillers based on computer vision was proposed. Firstly, the SEM images were preprocessed to remove noise and other factors affecting the statistics of particle size. Then, an improved watershed algorithm was used to segment the image. The particle size distribution of paper filler particles was determined by fast connected component labeling and Euler number algorithm. Finally, the result of particle size distribution from computer vision method was compared with those obtained by laser diffraction method. Experimental results of PCC paper filler images magnified 1500 times demonstrated that the particle size distribution of paper filler based on computer vision method was basically consistent with that measured by laser diffraction method with an error ratio less than 1% on median diameter. Moreover, the error ratio was about 6% when the cumulative particle size distribution reaches 85%.