摘要
为了实现图像处理技术对小麦容重影响因素的分析和容重的准确识别,研究了一种基于小麦图像特征和模式识别的小麦容重检测方法。采集不同容重小麦完整籽粒和籽粒横切面图像,对图像进行中值滤波、形态学运算、图像分割等处理,提取原图像与处理后图像的形态、颜色和纹理共3大类44个特征参数。最后采用逐步判别分析对提取的特征参数进行筛选,建立线性参数统计分类器和BP神经网络模型实现小麦不同容重的检测。结果表明,与小麦横切面图像特征相比,小麦完整籽粒图像的特征参数能更好的反映不同容重的差异;2种分类器对基于完整籽粒图像的小麦容重整体识别率均在95%以上。研究结果证明将图像处理技术应用于小麦容重检测识别是可行的。
In order to analyze the factors that affect the volume- weight and realize accurate identification of volume- weight of wheat by image processing technology,a novel detection method has been researched based on image features of wheat and pattern recognition method. Different volume- weights of wheat have been utilized and full grain images were taken as well as grain cross- section images. Then some image processing like median filtering,morphological operations and image segmentation were performed to extract 44 parameters from the three characteristic categories（ shape,color and texture）. STEPDISC was used to select image features and linear- function parametric statistical classifier. Back propagation neural network model was established to detect different volume- weights of wheat. The results showed that the features of full grain images expressed a more significant difference between different volume- weight than the features of grain cross- section images. The overall recognition rate of over 95% was achieved for full grain images using the statistical classifier and BPNN model. The results indicated that it was feasible to identify the volume- weight of wheat by image processing.
出处
《中国粮油学报》
EI
CAS
CSCD
北大核心
2015年第3期116-121,共6页
Journal of the Chinese Cereals and Oils Association
基金
国家自然科学基金(31371852)
河南工业大学校科学研究基金研究生教育创新计划(2012YJCX29)
关键词
图像处理
小麦
容重
逐步判别分析
神经网络
识别
image processing
wheat
volume-weight
stepwise discriminant analysis
neural network
identification
