中文摘要：

利用机器视觉判别农作物品种的应用日趋增多,为了提高粮食品种的识别率,提出利用稀疏表示的方法识别小麦品种。首先选择内乡188、郑麦9023、优展1号、豫麦47这4种不同类型的小麦品种,每种小麦随机选取200粒,选择其中40粒作为训练样本,提取可以区别不同类型小麦的颜色、形态和纹理特征参数作为小麦籽粒的典型特征,用来组建稀疏表示方法的所需字典;然后选择其中一种小麦品种作为测试样本,通过Matlab仿真计算每一个测试样本在字典上的投影,将投影误差最小的类作为测试样本的所属类别;最后对比和分析稀疏表示方法与BP神经网络方法对小麦品种的识别结果。仿真表明,稀疏表示方法对于4种小麦品种识别准确率达到96.7%,获得了很好的分类效果,是一种可以准确识别小麦品种的有效方法。

英文摘要：

The applications of using the machine vision to distinguish crop varieties grow increasingly,in order to improve the recognition rate of grain varieties,this paper proposes the method of using the sparse representation to identify the wheat varieties. First,for four different types of wheat varieties： Neixiang 188,Zhengmai 9023,Youzhan 1,Yumai 47,each of them randomly selected 200 wheat kernels,40 wheat kernels as the training sample,extracting the color,the shape and the texture feature parameters： which can distinguish different types of wheat varieties as the typical characteristics of wheat kernels,are used to form the dictionary of the sparse representation method. Then take one of the wheat varieties as test samples,calculate the projection of each test sample in the dictionary through Matlab simulation,and consider the minimum projection error as the class of the test sample. Finally,contrast and analyze the sparse representation method and the BP neural network method for the identification results of wheat varieties. The simulation shows that the identification accuracy of the sparse representation method for four wheat varieties can reach96. 7%,obtaining a very good classification effect,so the sparse representation method is effective which can accurately identify wheat varieties.