Abstract: Traditional wolfberry sorting primarily uses artificial method. It has time-consuming and inefficient shortcomings. An automatic wolfberry classification method based on machine vision is proposed. This paper uses digital image processing technology for wolfberry image pre-processing, segmentation and extraction of characteristic parameters of color, size and shape; it uses the K-means clustering feature to get the baseline of wolfberry appropriate level; it grades wolfberry by minimum distance classifier based on the trained benchmark. The experimental results show that this method can classify different colors and sizes of wolfberry more accurately and quickly.

Key words: wolfberry classification, image segmentation, feature extraction, cluster analysis, minimum distance classifier

摘要： 针对目前传统的枸杞分级主要采用人工方法，费时费力且效率不高的缺点，提出了一种基于机器视觉技术对枸杞进行自动分类的方法。采用数字图像处理技术对枸杞图像进行了预处理、分割，从而提取枸杞的色泽、大小及形状等特征参数；用K-means算法对特征进行聚类，得到枸杞相应等级的基准；根据聚类分析得到的基准采用最小距离分类器对枸杞进行分级。实验结果表明，该方法能够准确快速地对不同色泽和大小的枸杞进行分类。

关键词: 枸杞分级, 图像分割, 特征提取, 聚类分析, 最小距离分类器