基于图像处理和神经网络的小麦不完善粒识别方法研究  

Identification of unsound kernels in wheat based on image processing and neural network

作　　者：张玉荣[1] 陈赛赛[1] 周显青[1] 王伟宇[1] 吴琼[1] 王海荣[1] 

机构地区：[1]河南工业大学粮油食品学院,河南郑州450001

出　　处：《粮油食品科技》2014年第3期59-63,共5页Science and Technology of Cereals,Oils and Foods

基　　金：国家自然科学基金项目(31371852);河南工业大学校科学研究基金研究生教育创新计划(2012YJCX29)

摘　　要：为了实现图像处理技术对小麦不完善粒的准确快速识别,研究了一种基于小麦不完善粒图像特征和BP神经网络的不完善粒识别方法。采集小麦不完善粒图像,对图像进行中值滤波、形态学运算、图像分割等处理后,针对每个小麦籽粒,提取其形态、颜色和纹理共3大类54个特征参数,采用主成分分析法提取8个主成分得分向量作为模式识别的输入,建立BP神经网络模型,实现对小麦不完善粒的检测识别。结果表明,该模型对完善粒、破损粒、病斑粒、生芽粒和虫蚀粒的判别正确率分别为93%、98%、100%、90%和85%,平均判别正确率达到93%,可有效对小麦不完善粒进行检测识别。In order to identify accurately and fast the unsound kernels of wheat by image processing tech-nology, a novel detection method was studied based on image features of unsound kernels and BP neuralnetwork. The images of unsound kernels were captured and some image processings （median filtering,morphological operations and image segmentation etc. ） were performed to extract 54 parameters fromthree characteristic categories （shape, color and texture）. 8 principal components vectors were extractedas the inputs of pattern recognition by principal component analysis. The neural network model was estab-lished for identifying unsound kernels of wheat. The results showed that the recognition rate of sound ker-nels, broken kernels, spotted kernels, sprouted kernels and insect damaged kernels was 93% ,98%,100% ,90% and 85%, respectively, and the average recognition rate was 93%. It is concluded that thismethod is an effective way to identity unsound kernels of wheat.

关 键 词：图像处理 小麦 不完善粒 神经网络 识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TS210[自动化与计算机技术—计算机科学与技术]