融合HSV颜色空间和纹理特征的KNN 算法对马铃薯表皮的分类探索

打开文本图片集

关键词：马铃薯表皮分类；HSV 颜色空间；纹理特征；灰度共生矩阵；KNN 算法中图分类号：TP391.4 文献标识码：A 文章编号：2096-4706（2025）09-0025-07

Abstract： This paper realizes the accurate potato epidermis classification by using the KNN algorithm that combines the HSV color space and textural features. The collected potato epidermis images are converted from the RGB color space to the HSV color space， and the histograms of the hue， saturation， and brightness channels are calculated respectively. In terms of textural features， the GLCM （Gray Level Co-occurrence Matrix） is adopted to convert the images into grayscale images， and four feature values， namely energy， contrast， correlation， and homogeneity of the GLCM are calculated. Moreover， the LBP （Local Binary Pattern） histogram of the images is calculated through the LBP mode. Finally， these three kinds of features are concatenated into a one-dimensional array as the feature vector of the image， and the KNN （K-Nearest Neighbors） algorithm is utilized to achieve the potato epidermis classification. The results show that the accuracy is 95% ， the average precision rate is 91.67% ， the average recall rate is 95% ， and the average F1-score is 92.22% . The research indicates that the combined features can significantly improve the classification accuracy， providing a new and effective approach for the quality detection and classification of potatoes.

Keywords： potato epidermis classification; HSV color space; textural feature; Gray Level Co-occurrence Matrix; KNN algorithm

0 引 言

马铃薯是世界第四大粮食作物，在保障粮食安全和实现千年发展目标方面具有不可替代的作用，它的外部品质决定了其附加经济效益，高效准确的外部品质分级可将经济效益最大化 [1-2]。（剩余6982字）