基于活跃度的脉冲耦合神经网络的苹果腐烂分割方法  

Apple Rotten Area Segmentation Method Based on Activity of Pulse-coupled Neural Network

作　　者：邓晨曦[1] 周国雄[2] DENG Chen-xi;ZHOU Guo-xiong(Hunan Polytechnic of Environment and Biology,Hengyang 421005,Hunan;Central South University of Forestry and Technology,Changsha 410004,Hunan)

机构地区：[1]湖南环境生物职业技术学院,湖南衡阳421005 [2]中南林业科技大学,湖南长沙410004

出　　处：《湖南工业职业技术学院学报》2022年第1期9-13,19,共6页Journal of Hunan Industry Polytechnic

基　　金：2021年度湖南省教育厅科学研究项目“基于深度学习的语义级中文自动校对关键技术研究”(项目编号:21C1118)。

摘　　要：基于现有的机器视觉技术,针对水果的腐烂检测受到水果形状、大小、颜色影响而提取困难的问题,以苹果为研究对象,在机械自动化采摘时提出基于信息融合技术的苹果腐烂识别检测方法,检测苹果腐烂区域以便后续进行自动化处理。为了较好地分割苹果腐烂部分,提出一种基于活跃度的脉冲耦合神经网络图像分割算法,选择采用两个位置像素的联合概率密度定义的灰度共生矩阵,不仅描述了像素灰度信息,还描述了灰度空间分布信息,可以有效用来度量图像的复杂程度,较好地实现苹果腐烂分割。应用结果表明,采用改进的脉冲耦合神经网络图像分割算法较好地实现了苹果腐烂区域检测。In view of the problem that the existing machine vision technology for fruit rotten area detection is affected by the shape, size and color of fruits, taking apple as the research object, this paper proposes an apple rotten area identification and detection method based on information fusion technology in mechanical automatic picking, which detects the apple rotten areas for subsequent automatic processing. In order to segment the rotten parts of apples better, this paper proposes an image segmentation algorithm based on activity of pulse-coupled neural network. The gray level co-occurrence matrix defined by the joint probability density of two position pixels is selected to describe not only the gray level information of pixels, but also the spatial distribution information of gray level. It can be used to measure the complexity of the image effectively, and achieve better segmentation of apple rotten areas. The application results show that the improved image segmentation algorithm based on pulse-coupled neural network can detect the rotten areas of apple better.

关 键 词：活跃度 脉冲耦合神经网络 苹果腐烂区域检测 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]