摘要
针对苹果分级需要确定苹果正面的问题,本文结合简化的IDW算法,确定平滑处理苹果的方阵,与苹果图像进行卷积运算,通过两次平滑处理,识别出果梗。与常见的OSTU算法识别果梗对比,本文提出的算法识别效果更好。然后计算出果梗与苹果图像中心点的距离,与设定的阈值相比较,确定此苹果图像是否为苹果正面图像。试验表明,本文提出的算法,简单易操作,苹果正面识别正确率达97.5%。该方法不受苹果种类、大小等限制,具有较大的适用性,为苹果智能无损精准分级提供了依据。
As grading results of apples based on the feature of front shape are difficult to accurate,this paper combines the simplified IDW algorithm to determine the square matrix of the smoothed apple,and carries out convolution operation with the apple stem-end,compared with the OSTU,the aigorithm proposed in this paper has better recognition effect.Then the distance between the apple stem-end and the center of the apple image is calculated,and compared with the threshold value to determine whether the apple image is a positive image.The experiment shows that the method is simple and easy to operate,and the positive recognition accuracy of apple is up to 97.5%.The method is not limited by the kind and size of apple,and has a great applicability,which provides a basis for apple intelligent non-destructive accurate classification.
作者
王守亚
沈晓波
WANG Shouya;SHEN Xiaobo(School of Electronic Engineering,Huainan Normal University,Huainan 232038,China)
出处
《信阳农林学院学报》
2022年第1期110-114,共5页
Journal of Xinyang Agriculture and Forestry University
基金
2021年淮南师范学院校级科研项目(2021XJZD024)
2021年淮南市指导性科技计划项目(2021043)。
关键词
IDW算法
卷积运算
果梗
分级
IDW algorithm
Convolution operation
Apple Stem-end
Classificatio
