基于遗传神经网络的玉米叶色的自动测定研究

Investigation on Automated Color Measurement of Corn Leaves Based on Genetic Neural Network

摘要: 利用计算机图像处理技术和遗传神经网络技术，建立了一个多层前馈神经网络，实现了大田玉米和背景图像的正确识别，并且通过获取玉米叶的色度直方图提取了玉米叶表面颜色特征，进而求得了玉米叶色的测定值。实验结果表明，玉米叶色值自动测定系统，识别玉米的准确率可达91.6%，可以有效地测定玉米的叶色。该研究为实现大田玉米的化肥精确施用提供了理论依据。

Abstract: In this study, using image processing technology and genetic neural network technology, a three-layer feed forward neural network was established, which can identify corn from background correctly. The external color feature of corn leaves was extracted from the hue histograms. The measured color value of corn leaves was calculated. The experiment showed that corn could be recognized correctly by using this automated measurement system of corn leaf color value, the judging accuracy could attain 91.6%, and corn leaf color value could also be calculated correctly. This research work may provide theoretical foundation for precisely fertilizer spread in corn field.