基于计算机视觉的珍珠形状分级识别技术研究  

Separation method for shape grading of pearls using computer vision

作　　者：郑华文[1] 曹衍龙[1] 杨将新[1] 

机构地区：[1]浙江大学现代制造工程研究所,浙江杭州310027

出　　处：《工程设计学报》2008年第5期365-368,386,共5页Chinese Journal of Engineering Design

基　　金：国家高技术研究发展计划(863计划)资助项目(2007AA04Z426)

摘　　要：为了解决人工分选珍珠效率低、精度不高等问题,提出了一种基于计算机视觉的珍珠形状分级方法.首先根据珍珠图像的灰度直方图对图像进行分割,并使用闭操作去除图像中噪点,接着用八邻域法搜索珍珠图像的边界,提取珍珠轮廓曲线的傅立叶描述子,再计算被测珍珠与标准模板库珍珠的傅立叶描述子的欧氏距离,根据欧氏距离实现珍珠形状的分级识别.最后实验验证了本方法的有效性和实用性.A method based on computer vision for recognizing the pearls＇ shape is presented to solve the problem of low productivity and low accuracy of peal sort by hand. Firstly, the preprocessed color image is segmented according to its＇ grey degree histogram. Then the morphological closing operation is adopted to eliminate the noise points in the image. An 8-directional boundarytracing algorithm is applied to obtain the sequences of the boundary to abstract the Fourier De scriptorof peal＇s boundary curve. And then the Euclidean distance between the peals under checking and the ones in the standard template library, which is utilized to distinguish the pearls＇ shape. Finally the experimental result demonstrates the validity and practicability of this method.

关 键 词：珍珠分选 形状识别 傅立叶描述子 计算机视觉 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]