引用本文:[点击复制]
[点击复制]
【打印本页】【下载PDF全文】 查看/发表评论下载PDF阅读器关闭

←前一篇|后一篇→

过刊浏览    高级检索

本文已被:浏览次   下载 本文二维码信息
码上扫一扫!
基于计算机视觉技术的砀山酥梨果梗识别
徐旭艳,周平,吕冬,徐冯黎
0
(安徽农业大学工学院,合肥 230036)
摘要:
针对酥梨果梗计算机视觉检测方法的普适性问题,提出了一种利用像素点分析法识别果梗的算法。将酥梨图像进行二值化、滤波、形态学运算等预处理后,用Sobel算子提取酥梨图像边缘,再由一阶矩特征计算图像形心。选定形心为参考原点,建立图像的笛卡尔坐标系;从Y轴负方向开始,逆时针分别按15°和10°形心角对轮廓边缘点进行区域分块;统计分块区域内的像素点特征,并依此特征识别果梗。结果表明,对于无果梗的酥梨,15°和10°均能精确识别出来;有果梗的酥梨,15°较10°的识别率高,且含果梗区域的像素特征较其他区域的有很大突变。总体上果梗识别正确率可达97%,一定程度上可以满足酥梨果梗识别的需要。
关键词:  酥梨  果梗  像素畸变  像素点分析
DOI:10.13610/j.cnki.1672-352x.20150626.012
投稿时间:2015-02-02
基金项目:安徽农业大学人才稳定项目和校青年基金重点项目共同资助。
Recognition of the fruit stem of Dangshan pear based on the computer vision
XU Xuyan,ZHOU Ping,LV Dong,XU Fenni
(School of Engineering, Anhui Agricultural University, Hefei 230036)
Abstract:
In this work, a universal and applicable method based on algorithm of pixel analysis was proposed to identify the pear fruit stem. After binaryzation, filtration, and morphological operation, the image edge and the centroid of the pear fruit were extracted using Sobel operator and first-order moment. The centroid was set as the reference point. Additionally, a cartesian coordinate originated from the centroid was established. Then, the image edge was equally separated into domains by 15 degrees and 10 degrees from -90 degrees anticlockwise. Finally, the fruit stem was identified based on the calculated number of pixels in different regions. The results showed that the method of 15 degrees and 10 degrees can accurately identify the pear fruit that has no fruit stem; however, the method of 15 degrees was more accurate than the method of 10 degrees. Compared with the pixel's feature of other regions, a significant difference was observed in the zone of fruit stem. The overall accuracy of this method reached to 97%, suggesting a relatively applicable and sensitive strategy for recognition of pear fruit stem.
Key words:  pears  stem  pixel distortion  pixel analysis

用微信扫一扫

用微信扫一扫