光调制解调技术在田间杂草光谱识别系统中的应用研究

发布时间：2018-06-16 19:26

本文选题：杂草识别 + 光谱分析　；参考：《江苏大学》2017年硕士论文

【摘要】：杂草是一种人类不会主动培育而自由生长的草本植物。在中国种植面积较大的农作物中,油菜受草害尤为严重,在一般年份杂草大概会导致油菜减产10%到20%,而在一些极端条件下由于杂草的影响油菜甚至减产达到50%以上。鉴于杂草对农作物产生的严重危害,在田间往往需要对其进行主动防治,而喷施化学除草剂由于其操作简单且对杂草的防除效果好得到广泛的应用,然而,研究学者在除草剂过去许多年的使用中发现除草剂粗放式的大面积均匀喷施会对生态环境和人体健康会产生严重的不良影响。因此,具有空间选择性的对靶变量喷施方法成为目前最理想的除草剂施用方式,而实现对靶喷施的前提条件是解决对杂草的自动识别问题。基于光谱分析原理的传感器由于其检测实时性好、对样品无破坏、仪器成本相对较低等优势成为杂草自动识别领域的研究热点,并且已有基于此技术的传感器实现商业应用。但因为传感器本身组成原理方面的原因,传感器易受环境杂散光的干扰而降低其识别精度。光调制解调技术可以通过对光源发光强度进行调制并基于离散傅里叶变换进行解调剔除环境杂散光叠加在检测信号上的直流分量,从而减少环境杂散光对光谱检测仪器检测精度的影响。本文以冬油菜苗期杂草为检测对象,在项目组前期研究的基础上,使用590nm、710nm、750nm和940nm这四个检测波长,采用光调制解调技术,搭建了一套田间杂草光谱识别试验系统,并对其抗环境杂散光的能力进行了试验研究。主要研究内容如下:1.杂草识别光谱传感器识别精度影响因素分析:分别分析了传感器自身因素和外界环境干扰因素。传感器自身因素主要为光学系统设计的不合理、光电器件工作过程中产生的噪声以及信号采集处理系统在采样过程中产生的误差;外界环境干扰因素主要为天空背景光以及人造光源在传感器工作时对其产生的干扰。最后确定了外界环境杂散光是对测量精度产生影响的主要因素。2.基于光调制解调的田间杂草光谱识别试验系统的搭建:对光调制技术进行了简要介绍,对光信号的调制在抑制外界杂散光干扰和消除探测器自身产生噪声方面的作用进行了说明。运用光调制解调技术基于优选的四个特征波长(590nm、710nm、750nm和940nm)设计了一套实验系统,这套实验系统主要包括实验设备(光信号调制设备和光电信号采集设备)和实验数据处理LabVIEW程序。实验系统采用LED主动光源并对光源发光强度进行调制,采用基于光电二极管的增益可调的集成光电探测器检测反射光信号强度,采用工控机实时对探测器信号进行AD采样。用LabVIEW编写了实验数据处理程序,程序主要完成波形数据采样、波形数据收集、波形的数字滤波、滤波后波形的频谱测量和调制频率处信号幅值(即被测样品反射率表征值)的提取。3.光调制解调对环境杂散光干扰的剔除效果试验:为了验证应用了光调制解调技术的实验系统在剔除环境杂散光方面的效果,进行了两组验证试验,试验结果表明在外界环境杂散光缓慢变化和剧烈变化时,实验系统都能得到稳定的反射率表征值,这说明反射率表征值不受环境杂散光变化影响,光调制解调技术的应用极大的提升了测量的信噪比,为提高基于光谱原理的传感器测量精度提供了新的思路。4.实验系统静态性能试验:基于搭建的实验系统进行了静态性能试验,包括系统重复性测试、与光谱仪之间的标定试验及其验证试验。实验系统重复性测试结果的变异系数均小于1.5%,这表明实验系统的测试结果的一致性较好,实验系统稳定性较好,抗干扰能力强。对实验系统进行了标定试验,基于不同波长拟合出的四个拟合方程的决定系数分别为0.8692、0.9817、0.9812、0.9812,并以冬油菜为样本进行了验证试验,试验结果的相对误差都在10%以内,最大为8.82%,其中750nm和940nm的试验结果相对误差较小,大部分都在3%以内,最大为3.08%。从试验结果可以看出测得的反射率与被测样品的实际光谱反射率密切相关,可以通过标定后换算得到剔除了环境杂散光干扰的实际光谱反射率值。
[Abstract]:Weeds are a kind of herbaceous plants that are free to grow and grow freely. In large crops grown in China, rape is particularly severely damaged by grass. In general years, weeds will probably reduce the yield of rape by 10% to 20%, and in some extreme conditions, the effect of weeds on rape is more than 50%. The serious harm produced by crops is often required to be controlled actively in the field, and the spray chemical herbicide is widely used because of its simple operation and good effect on the prevention of weeds. However, researchers have found that the large area and uniform spraying of herbicides in the past many years in the use of herbicides will be applied to the ecological environment and the ecological environment. Human health will have serious adverse effects. Therefore, the spatial selectivity of the target variable spraying method has become the most ideal method of applying the herbicide, and the prerequisite for the implementation of the target spraying is to solve the problem of automatic identification of the weeds. The sensor based on the principle of spectral analysis is not able to break the sample because of its good real-time detection. The relative low cost of instrument has become a hot spot in the field of automatic weed identification, and the sensor based on this technology has been used for commercial applications. But because of the principle of the composition of the sensor, the sensor is susceptible to the interference of the ambient stray light to reduce its recognition accuracy. The luminescence intensity is modulated and based on the discrete Fourier transform to remove the DC component of the ambient stray light superimposed on the detected signal, thus reducing the influence of the environmental stray light on the detection precision of the spectral detection instrument. In this paper, the weeds at the seedling stage of the winter rape were tested on the basis of the preliminary study of the item group, using 590nm, 710nm, 7. The four detection wavelengths of 50nm and 940nm, using the optical modulation and demodulation technology, set up a set of field weed spectral identification test system, and carried out an experimental study on its ability to resist the environmental stray light. The main contents are as follows: 1. the influence factor analysis of the recognition accuracy of the weed identification spectral sensor: the factors of the sensor itself and the outside of the sensor are analyzed. The main factors are the unreasonable design of the optical system, the noise produced in the process of the optoelectronic devices and the error produced by the signal acquisition and processing system in the sampling process, and the interference factors of the external environment are mainly the background light of the sky and the interference of the artificial light source to the sensor when working in the sensor. Finally, the external ambient stray light is the main factor affecting the measurement accuracy,.2. based on the optical modulation and demodulation in the field of the field weed spectrum identification test system. The light modulation technology is briefly introduced, the modulation of the light signal is used to suppress the external stray light interference and eliminate the noise of the detector itself. The optical modulation and demodulation technology is based on the optimum four characteristic wavelengths (590nm, 710nm, 750nm and 940nm) to design an experimental system. The experimental system mainly includes the experimental equipment (optical signal modulation equipment and photoelectric signal acquisition equipment) and the experimental data processing LabVIEW program. The experimental system uses the LED active light source and the light. The intensity of the source luminescence is modulated. The integrated photoelectric detector based on the gain of photodiode is used to detect the intensity of the reflected light signal. The AD sample is sampled in real time by the industrial control machine. The experimental data processing program is written in LabVIEW. The program is mainly used to complete the data sampling, the waveform data collection and the digital filtering of the waveform. The spectrum measurement after filtering and the amplitude of the modulated frequency at the modulated frequency (i.e. the measured value of the measured sample reflectivity) the elimination effect of.3. light modulation and demodulation on the ambient stray light interference: in order to verify the effect of the experimental system using the optical modulation and demodulation technology in eliminating the environmental stray light, two sets of verification tests are carried out. The results show that the experimental system can obtain a stable reflectance characterization value when the external ambient stray light changes slowly and violently, which indicates that the reflectance characterization value is not affected by the change of ambient stray light. The application of the optical modulation and demodulation technology greatly improves the signal to noise ratio of the measurement, and improves the measurement precision of the sensor based on the principle of spectrum. The static performance test of the new.4. experimental system is provided: the static performance test based on the built experiment system, including the system repeatability test, the calibration test and the verification test between the spectrometer and the spectrometer. The coefficient of variation of the experimental system repeatability test results is less than 1.5%, which indicates the consistency of the test results of the experimental system. The experimental system has good stability and strong anti-interference ability. The calibration test of the experimental system is carried out. The determination coefficients of four fitting equations based on different wavelengths are 0.8692,0.9817,0.9812,0.9812 respectively. The test results are tested with winter rape as samples, and the relative error of the test results is within 10% and the maximum is 8.82%. The relative error of the experimental results of 750nm and 940nm is relatively small, most of which are within 3%. The maximum of 3.08%. is that the measured reflectance is closely related to the actual spectral reflectance of the measured samples. The actual spectral reflectance of the ambient stray light interference can be obtained by the calibration.
【学位授予单位】：江苏大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212;S451

【参考文献】