论文部分内容阅读
以南疆红枣颜色分级为研究对象,从预处理后的红枣图像中提取红体均值(R)、绿体均值(G)和蓝体均值(B)以及它们的均方差σR、σG、σB共6种颜色特征变量;再将图像从RGB到HIS颜色空间转换,然后从HIS颜色空间中,提取色度均值(H)、亮度均值(I)和饱和度均值(S)以及它们的各自的均方差σH、σS、σI共6个颜色特征变量,总计12个颜色特征变量,最后应用BP神经网络进行红枣颜色分级。结果表明,BP人工神经网络分级与人工分级的一致度达到了91.6%,该网络分级效果良好,能较好地满足红枣颜色分级的需求,对南疆红枣产品的生产、销售具有一定的理论指导和实际应用意义。
Taking the color grading of red jujube in southern Xinjiang as the research object, the red body mean (R), the green body mean (G) and the blue body mean (B) and their mean square deviations σR, σG and σB were extracted from the pretreated jujube images 6 color feature variables; and then converting the image from RGB to HIS color space, and then extracting the hue mean (H), the mean brightness (I) and the mean saturation (S) from the HIS color space and their respective mean Variance σH, σS, σI a total of six color characteristic variables, a total of 12 color characteristic variables, and finally the application of BP neural network for jujube color grading. The results show that the degree of agreement between BP artificial neural network grading and artificial grading reached 91.6%. The network grading effect is good and can better meet the needs of color classification of jujube, and has certain theoretical guidance on the production and sales of red jujube in southern Xinjiang And practical significance.