基于图像处理技术的芸豆盐碱胁迫检测方法

Method for Detecting Salt Alkali Stress in Kidney Beans Based on Image Processing Technology

为解决传统的芸豆盐碱胁迫检测方法存在仪器设备繁多、操作复杂和化学试剂对植物有损害等问题,研究建立了基于图像处理技术的芸豆盐碱胁迫的快速无损检测方法。通过人工培育不同盐碱胁迫度的芸豆样本,采集其冠层彩色图像和热红外图像,并提取出图像中的冠层区域,利用图像处理技术计算彩色图像、热红外图像的纹理特征和频谱特征,以获取芸豆冠层异构图像的70维度特征为基础,构建RBF模型的网络结构为70-22-6型,实现芸豆盐碱胁迫度分类检测的平均准确率和仿真时间为86.5%和0.71 s。该模型是一种具有高效准确性,易推广的芸豆盐碱胁迫检测的新方法,能够为快速无损地检测豆类作物的盐碱胁迫程度提供技术支撑。

In order to solve the problems of the traditional methods of kidney bean salt and alkali stress detection,such as many instruments,complex operation and chemical agents damage to plants,a rapid non-destructive detection method based on image processing technology was developed.By artificially cultivating kidney bean samples with different salinity and alkali stress,canopy color images and thermal infrared images were collected,and the canopy region in the images was extracted.Image processing technology was used to calculate the texture and spectrum characteristics of color images and thermal infrared images,and on the basis of obtaining the 70-dimension characteristic vector of heterogeneous kidney bean canopy images.The network structure of the RBF model was 70-22-6,and the average accuracy and simulation time of salt and alkali stress classification of kidney bean were 86.5%and 0.71s.This model was an efficient,accurate and easy to popularize new method for the detection of salt and alkali stress in kidney beans,which could provide technical support for the rapid and non-destructive detection of salt and alkali stress degree of bean crops.