李亚文^1，2*，陈月星³，呼高翔¹

（1.商洛学院电子信息与电气工程学院，商洛 726000；2.商洛市人工智能研究中心，商洛 726000；3.商洛学院生物医药与食品工程学院，商洛 726000）

摘要：准确识别与防治苹果叶部病害，能够有效提高苹果的产量与品质。以常见的苹果叶部病害(锈病、黑腐病、黑星病)为研究对象，构建基于KNN和多特征融合的无损检测模型。使用K-means聚类算法分割苹果叶部图像，通过颜色矩、灰度共生矩阵、Hu距分别提取图像的颜色、纹理和形状特征，利用KNN对特征参数进行分类模型训练，能够实现绿色准确识别苹果叶部病害的目的。实验结果表明，以颜色、纹理、形状为单特征检测的苹果叶部病害识别精确率分别为75%、57%、45%，其中颜色特征更加直观，有9个特征量识别率较高，形状特征在进行图像分割时很难确定K点导致识别率低。该研究基于颜色、纹理、形状等多特征融合提取13个特征量，能够准确识别苹果叶部病害，其识别率达84%，为实现绿色农业果园病虫害防治提供技术支持。

关键词：K-近邻方法；K-means聚类算法；多特征融合提取；苹果叶部；病害识别

中图分类号：TP391.41；S436.611 文献标识码：A 文章编号：1674-506X（2024）04-0025-0008

Detection of Apple Leaf Disease Recognition Algorithm Based on KNN and Multi-feature Fusion

LI Yawe^n1，2*，CHEN Yuexing³，HU Gaoxiang¹

（1.College of Electronic Information and Electrical Engineering，Shangluo University，Shangluo 726000，China；2.Research Center for Artificial Intelligence of Shangluo，Shangluo 726000，China；3.College of Biomedical and Food Engineering，Shangluo University，Shangluo 726000，China）

Abstract：Accurate identification and prevention of apple leaf diseases can effectively improve the yield and quality of apples. As the research object of the common apple leaf diseases (rust，black rot and scab)，A nondestructive detection model based on KNN and multi- feature fusion is constructed. The K-means clustering algorithm was used to segment the apple leaf image. The color，texture and shape features of the image were extracted by color moment， gray level co-occurrence matrix and Hu distance respectively. The characteristic parameters were trained of the classification model by the KNN algorithm， which can realize the purpose of green and accurate identification of apple leaf diseases. The experimental results showed that the accuracy of apple leaf disease recognition based on single feature detection of color，texture，and shape was 75%，57%，and 45%，respectively. The color feature is more intuitive with 9 features，and the recognition rate is higher. The shape feature is difficult to determine the K point when performing image segmentation，resulting in a low recognition rate. Based on color，texture，shape and other multi-feature fusion，13 features were extracted，It can accurately identify apple leaf diseases with a recognition rate of 84%， and provides technical support for the prevention and control of pests and diseases in green agricultural orchards.

Keywords：K-nearest neighbor method；K-means clustering algorithm；multi-feature fusion extraction；apple leaf；disease identification

doi：10.3969/j.issn.1674-506X.2024.04-005

收稿日期：2024-03-13

基金项目：陕西省科技厅科技计划项目（2023-JC-QN-0661）；商洛学院科研创新团队（19SXC03）；陕西省本科高等教育教学改革项目（21BY162）

作者简介：李亚文（1984-），女，副教授。研究方向：农业信息化、机器视觉与模式识别、目标检测与跟踪研究。

*通信作者

引用格式：李亚文，陈月星，呼高翔.基于KNN和多特征融合的苹果叶部病害识别检测［J］.食品与发酵科技，2024，60（4）：25-32.