摘要
主要探讨脱绒棉种颜色特征与种子活力之间的相关性,搭建脱绒棉种图像采集平台,并进行图像处理算法的研究。在RGB、HSV和I1I2I3颜色模型下提取颜色特征参数,同时进行脱绒棉种电导率的测定,采用SPSS19对各颜色特征参数与脱绒棉种电导率值进行相关性分析,结果表明脱绒棉种部分外观颜色特征参数与其电导率值之间相关性显著。基于BP神经网络的脱绒棉种内部品质检测模型,取脱绒棉种鼎丰10和新陆早45各300粒对模型进行训练,当鼎丰10和新陆早45隐含层的结点数分别选择为8和7时,网络均方误差和函数收敛效果较好,然后将脱绒棉种鼎丰10和新陆早45各取100粒对训练好的网络模型进行验证,结果表明用BP神经网络的检测精度分别可达到82.7%、86.1%。
The correlation between color features and seed vigor of cottonseed was discussed.The platform of the cottonseeds image acquisition was built,and the image processing algorithm was studied.The cottonseeds＇images were analyzed under the RGB,HSV and I1I2I3 color model,and cottonseeds conductivity was measured simultaneously.The results of the correlation analysis about each color characteristic parameters and cottonseeds conductivity value by using SPSS19 showed that some cotton color characteristic parameters and conductivity value presented highly correlated.Using the internal quality testing model based on the BP neural network,three hundreds of cottonseeds named Ding Feng 10 and New Land Early 45 were put in training the model respectively.When the hidden layer nodes of Ding Feng 10 and New Land Early 45 were 8and 7,the network MSE and function convergence were better.Then the two cotton varieties of 100 grains were selected respectively to verify the experiment.The rates of detection using BP neural network can reach 82.7% and 86.1% respectively.
出处
《中国农机化学报》
2015年第4期111-114,共4页
Journal of Chinese Agricultural Mechanization
基金
国家自然科学基金(31260290)--基于外观信息感知的脱绒棉种内部品质检测机理研究
