基于虚拟仪器技术的田间多光谱视觉系统设计

In-field Multi-spectral Computer Vision System Design Based on Virtual Instrument Technology

设计了一套田间多光谱虚拟仪器视觉系统。系统使用高分辨率的多光谱(近红外、红光和绿光)相机MS3100,拍摄作物生长期的多光谱图像,采用Labview及其视觉模块编写图像的采集、处理和分析程序,实时测取作物各个光谱波段的反射率。田间试验表明,该系统可以准确地对图像中的作物进行识别,求取作物的光谱反射特征,在2.4 m×1.8 m的视窗内,每组图像的采集和处理时间平均为311 ms,满足田间精准变量投入的在线工作要求。

An in-field multi-spectral computer vision system was developed using virtual instrument for capturing crop growth status precisely in real time to supervise a variable rate application. The system used a high resolution multi-spectral camera （MS3100） of acquiring excellent multi-spectral images （near-infrared, red and green） to shoot crop images. Labview and its vision development module were adopted to do image acquisition, processing and analysis programs for measuring the crop reflectance of red, green and near-infrared. The test trail of system shows that it can analyze the crop multi-spectral images and compute the spectral characteristics correctly. The average running time of acquiring and processing images of each group （NIR, R and G） were 311 ms in the range of 2.4 m× 1.8 m. The running speed of system can satisfy the in-field on-line job of the variable precision application.