| 吴建清，苏信晨.基于机器视觉的芒果检测与分级研究[J].海南师范大学学报自科版,2024,37(1):56-64 |
| 基于机器视觉的芒果检测与分级研究 |
| Research on Mango Detection and Grading by Machine Vision |
| |
| DOI： |
| 中文关键词: 芒果 机器视觉 HALCON 分类器 |
| 英文关键词: mango machine vision HALCON classifier |
| 基金项目:海南省高等学校科学研究项目（Hjkj2013-23） |
|
| 摘要点击次数: 123 |
| 全文下载次数: 195 |
| 中文摘要: |
| 为了提高贵妃芒果检测与分级的准确率和效率，首先用标定好的工业相机对芒果进行拍照，然后使用HALCON对芒果图像进行灰度化和图像分割预处理，接着提取芒果面积、果形指数、成熟度、缺陷面积和缺陷个数5个特征参数并归一化，把它们分别作为GMM、MLP、SVM和KNN分类器的输入向量，并以芒果的4个等级作为分类器的输出向量，最后以每级120个训练样本，60个测试样本分别对4种分类器进行训练和测试。结果表明4种分类器的平均准确率依次为92.5%、93.75%、98.75%和98%，准确率较高，有一定的实际应用价值。 |
| 英文摘要: |
| In order to improve the accuracy and efficiency of mango detection and grading of Royal mango. Firstly, we take photos of mango with a calibrated industrial camera, the mango image is pre-processed with HALCON for graying and image segmentation. Five characteristic parameters of mango area, fruit shape index, maturity, defect area and defect number are extracted and normalized, then we take them as input vectors of GMM, MLP, SVM and KNN classifiers respectively and take the four grades of mango as output vectors of the classifier. Finally, 120 training samples and 60 test samples are used to train and test the four classifiers. The results show that the average accuracy rates of the four classifiers are 92.5%, 93.75%, 98.75% and 98% respectively. The accuracy rates are all high and have certain practical value. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
