基于计算机视觉的苹果水心病无损在线分级系统  

Lossless and Online Classification System for Apple Water Core Disease Based on Computer Vision

作　　者：张弘炀 蔡骋[1] Zhang Hongyang;Cai Cheng(College of Information Engineering, Northwest A＆F University, Yangling 712100, China)

机构地区：[1]西北农林科技大学信息工程学院,陕西杨凌712100

出　　处：《农机化研究》2018年第10期208-210,共3页Journal of Agricultural Mechanization Research

基　　金：陕西省农业科技创新与攻关项目(2015-Q1)

摘　　要：对苹果按照水心病患病程度进行无损在线分级对于苹果采摘机器人具有重要的意义。在苹果园中,利用无损在线检测出轻度患有水心病的苹果果实并优先采摘,以避免因患病苹果烂掉而造成经济损失,构建了基于计算机视觉和高光谱技术的苹果水心病患病程度无损在线分级系统。本系统以西北农林科技大学白水苹果试验示范站的秦冠苹果为研究对象,采集苹果果实的高光谱图像和果面图像,对苹果图像进行压缩感知预处理和计算机视觉特征提取,并使用支持向量机算法根据苹果图像的计算机视觉特征来对苹果的水心病患病程度进行分级。实验表明:该系统对苹果水心病的分级准确率可以达到78.2%。与人工对苹果水心病分级相比,该系统不依赖于农业专家对苹果水心病特征的丰富知识,不受农业专家的主观影响。本系统作为苹果采摘机器人的一个模块可以为苹果的采摘行为提供指导,降低了人工成本,提高了经济效益。The classification of apple water core disease is very important for apple-picking machine.In order to detect apples slightly infected by water core disease and pick them preferentially in apple orchard,build a classification system for apple water core disease based on computer vision and hyper-spectral.This research collect hyper-spectral pictures and surface picture in lossless and online manner for Qinguan apple in the Baishui experimental apple station,which belongs to Northwest A F University.After pre-processing,extract computer vision features for these apple pictures.Then classify the level of water core disease for these apple pictures using support vector machine.Experiments show that water core disease classification accuracy of this system can achieve 78.2 %.Compared with classify the level of water core disease manually,this system doesn＇t require agricultural expert with professional experience and is not subjective.This system can serve as a module of apple-picking machine,which can decrease the cost for labor and increase the profit.

关 键 词：苹果 水心病 采摘机器人 计算机视觉 

分 类 号：S24[农业科学—农业电气化与自动化] TP391.41[农业科学—农业工程]