基于深度颜色特征的灰度直方图玉米品种识别研究  

Study on Corn Variety Identification Based on Depth and Color Features of Gray Histogram

作　　者：张云丽[1] 韩宪忠[1] 王克俭[1] 

机构地区：[1]河北农业大学信息科学与技术学院,河北保定071001

出　　处：《作物杂志》2015年第1期156-159,F0003,共5页Crops

基　　金：河北农业大学理工科学研究基金(LG201106);河北省科技厅科技富民强县行动示范县建设专项(14227404D)

摘　　要：为了提高玉米品种识别的准确率,提出了一种基于深度和颜色的灰度直方图结合BP神经网络的玉米品种分类方法。使用深度传感器获取玉米子粒的深度图像,并将获得的RGB彩色图像转化为HSV图像进行分析,发现不同品种的H分量有明显差异,从而确定不同颜色范围对应的灰度值,用归一化和灰度化后的图片生成灰度直方图,发现不同品种的灰度特征值差异比较大,取其中重要的4个灰度特征值作为BP神经网络的输入,经过训练识别出不同的品种。试验结果表明,此方法识别出的玉米品种与人眼观察的结果基本一致。To enhance the accuracy of the corn variety identification, this paper proposes a method to recognize corn varieties based on depth and color of gray histogram combined with BP neural network. Range sensors was used to obtain the depth of the corn grain image and the RGB color images were transformed into HSV image for analysis. We found that different varieties of H component had obvious differences. With that we could determine the grey value of different color range accordingly. Then the gray-level histograms were constructed by the normalization and gray level. It was found that different varieties of gray characteristic value had remarkable differences. Four impor- tant gray characteristic values were employed in the BP neural network input which was trained to identify different varieties. The experimental results show that using this method to identify the corn varieties basically conforms to re- sults identified by human eyes.

关 键 词：深度玉米图像 HSV模型 灰度直方图 BP神经网络 品种识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]