UAV-based hyperspectral poppy rapid monitoring solution

2024-01-24

Preface

Drugs are like tumors that cause profound harm to the social system. What they produce is a negative economy. According to the principle of economics, carrying out drug control and reducing drug production is a process that generates social and economic benefits. Therefore, countries around the world all actively participate in anti-drug activities and combating drug production, processing, trafficking and drug abuse.

Poppy (Papaver somniferum L) is a kind of hypnotic poppy plant. It is the main ingredient for producing opium (pharmacologically active ingredients are present in its alkaloids, the most important of which is morphine, which accounts for about 10% of the weight of raw opium). Raw materials and extracts are also the source of a variety of narcotics, including morphine (and its derivatives heroin), thebaine, codeine, papaverine and noscapine. Therefore, poppy is not only dangerous for making drugs, but also has medicinal value and the value of producing poppy seeds.

In 2006, the area of opium poppy cultivation dropped significantly in many regions around the world, but there was also a substantial increase in some areas. The global opium poppy cultivation area increased by 31%, mainly distributed in Afghanistan (accounting for 93% of the world), Myanmar, South America, and North Korea. and other areas. The current anti-drug situation faced by our country is also very severe and complex. Drug crimes are frequent and high, and the anti-drug task is very arduous. Therefore, accurately identifying the original plant of drugs is the key to anti-drug work, and opium poppy, as a major original plant of drugs, is the focus of identification. The cultivation of illegal opium poppies is generally hidden, and traditional manual ground identification and investigation methods require a lot of manpower and material resources. The use of drone aerial photography technology to identify illegal poppies has the advantages of wide coverage, reliable information sources, and resource saving. Compared with panchromatic cameras, airborne hyperspectral cameras can obtain spectral reflectance information of hundreds of channels of ground objects, thereby accurately identifying specific poppies by comparing the differences in reflectance of different ground objects in different channels.

Figure 1 Diagram of different parts of poppy.

The current poppy detection and control methods based on manual on-site surveys and sample plot methods are no longer able to meet the requirements of the current intensified severe situation. With the evolution of earth observation platforms, the advancement of geographical information spatio-temporal analysis technology, the popularization of navigation and positioning technology, the innovation of machine learning algorithms, breakthroughs in computing power, and the accumulation and opening of crime and other spatio-temporal big data, significantly improve the role of space earth observation and analysis application technology in crime analysis, and promotes the development of crime analysis theories, methods and applications. Remote sensing technology can explore poppy cultivation conditions and spectral characteristic mechanisms from multiple angles and levels. Compared with traditional monitoring methods, its information is more comprehensive and accurate, and its conclusions are more convincing.

Since 2004, China's National Narcotics Control Office has officially used satellite remote sensing technology to conduct remote sensing identification of illegal and criminal activities of illegal opium poppy cultivation in mountainous forest lands in key areas where illegal cultivation exists across the country, improving the level of national ban on opium poppy cultivation and drug eradication work. and discovery capabilities. In 2013, low-altitude drones were also included in the monitoring system and used to detect sporadic illegal planting in plain areas, which improved the detection efficiency of illegal planting from the perspective of science and technology police. According to the "2015 China Drug Situation Report", through satellite remote sensing and drone methods, a total of 3.06 million opium poppies were eradicated from 289 acres across the country in 2015. The problem of illegal and sporadic opium poppy cultivation across the country has been repeated, but it has dropped by about 53% and 30% year-on-year. The "2016 China Drug Situation Report" shows that large-scale illegal cultivation of original drug plants in the country is basically prohibited. A total of 84 acres and 1.16 million opium poppy plants were found illegally planted across the country, a year-on-year decrease of 70.9% and 62%. Sporadic illegal cultivation problems still exist in some areas.

Technical ideas and main content

UAV hyperspectral images and fieldspec spectroradiometer are used to monitor the status of poppy cultivation to establish a complete poppy sample library. Poppy target detection is carried out through the current mainstream deep convolutional neural network model to determine the key areas for poppy cultivation, and then use fieldspec spectroradiometer collects sample libraries of opium poppies in different periods, and compares the sample libraries to divide the opium poppies in different periods to obtain the sensitive wavebands of opium poppies. Finally, the results are displayed through the monitoring management platform. The specific technical route is shown in Figure 2.

Figure 2 Poppy cultivation investigation technology flow chart.

What needs to be implemented:

1) Use a fieldspec spectroradiometer to collect spectral curves of poppy in different periods, establish a database of poppy samples, and then use correlation analysis to screen sensitive characteristic bands of poppy to form a poppy characteristic band library and provide more complete statistical information for monitoring the growth status of poppy in different periods.

2) Real-time monitoring of the distribution, area, and quantity of opium poppy in a certain area based on drone hyperspectral images, focusing on understanding the illegal cultivation of opium poppy and conducting investigations.

3) Realize linkage management. The management of the platform can be accurate to small classes, which facilitates task dispatch, task tracking and management of the anti-drug brigade. It improves the platform’s ability to view the “one picture” of poppy distribution and the analysis and display of poppy-related data (such as regional poppy distribution, quantity changes, anti-drug brigade grid management data, etc.).

4) The drone flies once a month. Based on the long-term sequence of monitoring image data, it can timely understand the distribution, area, and quantity changes of opium poppy planting in a county or city, thereby providing a reference for accurate monitoring.

Technical points

Original spectral curve of poppy

An important condition for building a poppy spectral curve library is to analyze the spectrum of the endmembers of the research object. Based on the poppy endmembers at different growth periods, 100 poppy samples at different growth periods were selected for fieldspec spectroradiometer data collection and statistical analysis, and the poppy spectral curves at different growth periods were extracted for correlation analysis and other operations to establish a poppy characteristic sample library. Correspond the spectral data collected by hyperspectral to the sample library established by the fieldspec spectroradiometer, and establish a hyperspectral characteristic band data set for use in the next step of target detection.

Figure 3 Poppy spectral curve.

2. Various index spectra of vegetation

In order to further widen the difference between poppy and other genera of plants, various spectral indices can be constructed using the original spectra, as shown in Table 1.

	Vegetation index name	Calculation formula
GI	Greenness index	R544/R677
SIPI	Structure insensitive pigment index	(R800-R445)/(R800-R680)
NPCI	Normalized total pigment chlorophyll index	(R680-R430)/(R680+R430)
MSR	Modified simple ratio index	(R800/R670-1)(R800/R670+1)^1/1
NRI	Nitrogen reflectance index	(R570-R670)/(R570+R670)
PRI	Photochemical reflectance index	(R570-R531)/(R570+R531)
TCARI	Transformed chlorophyll absorption in reflectance index	3[(R700-R670)-0.2(R700-R550)*(R700/R670)]
PSRI	Plant senescence reflectance index	(R800-R445)/(R800-R680)
PHRI	Physiological reflex index	(R550-R531)/(R550+R531)
ARI	Anthocyanin reflectance index	(R5300^(-1)-(R700)^(-1)
TVI	Triangle vegetation index	0.5[120(R750-R550)-200*(R670-R550)]
RVSI	Red edge vegetation stress index	(R712+R752)/2-R732
MCARI	Modified chlorophyll absorption reflectance index	[(R701-R671)-0.2*(R701-R549)]/(R701/R671)
ARVI	Atmospherically resistant vegetation index	[R800-2(R700-R436)]/[R800+2(R700-R436)]
DVI	Difference vegetation index	R800-R700
EVI	Enhanced vegetation index	2(R800-R700)/(R800+6R700-7.5*R436+1)
GNDVI	Greenness normalized difference vegetation index	(R546-R700)/(R546+R700)
LMI	Leaf moisture index	R1650/R830
OSAVI	Optimize soil adjustment vegetation index	[(R800-R700)/(R800+R700+0.16)]*(1+0.16)
NDVI	Normalized difference vegetation index	(R800-R700)/(R800+R700)
RVI	Ratio vegetation index	R800/R700
SAVI	Soil adjustment vegetation index	1.5*(R800-R700)(R800+R700+0.5)

Table 1

Program implementation

1. Hyperspectral drone flight service

Hyperspectral reflects the characteristics of high-resolution optical information, which uses many very narrow electromagnetic wave bands (usually <10nm) to obtain relevant data from objects of interest. Hyperspectral images are acquired by an imaging spectrometer, which provides tens to hundreds of narrow-band spectral information for each pixel, producing a complete and continuous spectral curve. Image information can reflect the external quality characteristics of the sample such as size, shape, defects, etc. Since different components have different spectral absorption, the image will have a more significant reflection on a certain defect at a specific wavelength, and the spectral information can fully reflect the sample differences in internal physical structure and chemical composition.

UAV hyperspectral has the following characteristics:

① Many spectral features: The imaging spectrometer has 300 bands in the visible and near-infrared spectral regions;

②High spectral resolution: The sampling interval of the imaging spectrometer is small, and the resolution is less than 3nm. Fine spectral resolution reflects the subtle characteristics of the spectrum of ground objects;

③ Rich amount of data: As the number of bands increases, the amount of data increases exponentially. It can provide spatial domain information and spectral domain information, that is, "unification of images and spectra", and the spectral curve obtained by the imaging spectrometer can be compared with the spectral curve of similar ground objects measured on the ground.

Comparison with visible light results

Visible light images only consist of light information in three bands: red, green, and blue (RGB). When identifying poppies, they are monitored through manual interpretation, which is inefficient and has limited accuracy. The hyperspectral image results have 300 bands of light information. For poppy, its exclusive band characteristics can be extracted to accurately identify it. At the same time, combined with machine learning/deep learning algorithms, automated monitoring can be achieved.

Comparison with multispectral results

The main difference between multispectral and hyperspectral is the number and narrowness of the bands. Multispectral images usually consist of 10-30 bands of light information. They have limited detection features for poppies or other ground objects and are easily confused with the reflection spectra of other ground objects. Having a higher level of spectral detail in hyperspectral images can provide better vegetation discrimination capabilities. For example, when detecting poppies, hyperspectral images can effectively distinguish poppies from other similar plants and improve accuracy.

Multispectral images usually refer to 10 to 30 bands expressed in pixels, and each band can be obtained by using a remote sensing radiometer.

Hyperspectral images contain very narrow bands (<10 nm), and hyperspectral images have hundreds of bands (for example, ATH9010 has 480 bands).

In the planned research area, hyperspectral images are collected according to needs. Specific steps are as follows:

(1) UAV side settings: Assemble the UAV hyperspectral equipment, set the altitude and speed, and set the image route spacing according to the camera parameters and image overlap requirements;

(2) Camera side settings: Set the camera frame rate according to the altitude and speed, and set the integration time (exposure time) according to the whiteboard measurement value;

(3) Standard reflectivity whiteboard: Place a standard reflectivity whiteboard in the route area, and the whiteboard needs to be captured when collecting images.

Figure 6 Schematic diagram of UAV hyperspectral data collection.

2. UAV hyperspectral image processing

After collecting UAV hyperspectral image data, the following preprocessing work is required:

(1) Wavelength calibration: The original image collected does not have wavelength information, and a wavelength calibration file needs to be added;

(2) Image cropping: Hyperspectral uses push-broom imaging, which requires cropping of the collected images of the measurement area;

(3) Registration and splicing: perform geo-registration or relative registration on the cropped survey area images, and then splice the registered images into a complete image;

(4) Radiation correction: The value in the original image represents the reflection intensity, which needs to be corrected using the whiteboard reflection value and standard reflectance to calculate the reflectance of the entire image.

(5) Mixed spectral decomposition: The spectral data collected by drones are affected by the spatial resolution of the image, resulting in a pixel that may be mixed and averaged by different ground objects or vegetation. In order to improve the accuracy, a mixed spectral decomposition operation needs to be performed on the image;

(6) Spectral filtering (smoothing): There is a certain amount of noise in the spectral information in the original image, and spectral filtering needs to be performed before application.

3. Collection of ground object sample data

In addition to UAVs collecting hyperspectral images of the study area, on-site surveys also need to be carried out using fieldspec spectroradiometer. The purpose is to provide real training samples of the on-site classification model and verification samples of the classification results for the classification of UAV remote sensing images. The main content of the on-site investigation is to measure the spectrum data of poppy at different growth stages. The spectrum data of poppy at different growth stages collected by the fieldspec spectroradiometer will be used as a standard training sample data set for UAV hyperspectral image data processing.

Figure 7 Schematic diagram of ground object spectrometer equipment and data collection.

4. Poppy target detection results

First, a fieldspec spectroradiometer is used to collect spectral data of poppy in different periods, and then statistical and correlation analysis is performed to determine the sensitive band of poppy. Then, corresponding to the hyperspectral band range, a deep learning training sample library is established to train the convolutional neural network algorithm. The specific results of automatic poppy identification are shown in the figure.