基于计算机视觉的红枣叶片病害严重度估测方法

对红枣叶片病害严重度的精准评估是采取有效防治的关键,而传统的评估方法费时费力。该研究以病斑严重度不同导致的颜色特征差异为依据,提出了一种基于计算机视觉的红枣叶片病害严重度估测方法。对叶片病斑图像预处理后,提取基于R、G、B颜色空间的8个特征向量作为模型的输入变量,以正常、轻微、一般和严重作为模型的输出,利用GA-BP神经网络建立了红枣叶片病害严重度的估测模型。实验结果表明,模型可实现对红枣叶片病害严重度的快速识别,识别精度达到了87%以上。

A Method of Estimating Red Jujube Blade Disease Severity Based on Computer Vision

The accuracy of estimating red jujube blade disease severity was the key to effective prevention,but the traditional evaluation method was slow and arduous.This research provides a method of estimating red jujube disease severity based on computer vision according to color differences of different disease severity.After blade disease spot image is pretreated,and to extract eight input characteristic vectores based on R,G,B color space,as the output of the model in normal,mild,general and serious,finally,using the BP neural network that the genetic algorithm have improved to establish estimation model of red jujube blade disease severity.Experimental results show that Model can quickly identify red jujube blade infection disease severity.The accuracy reaches more than 87 percent.