In recent 20 years, hyperspectral remote sensing technology has developed rapidly, which integrates detector technology, precision optics and machinery, weak signal detection, computer technology and information processing technology, and has become one of the frontier technologies in the field of remote sensing.
The origin and development of hyperspectral remote sensing 1.2. 1
With the continuous progress of basic theory and material science, hyperspectral remote sensing technology has developed rapidly in recent 20 years, and has become another important research direction in the field of remote sensing besides radar remote sensing, laser remote sensing and ultra-high resolution remote sensing.
1.2. 1. 1 development of hyperspectral imager abroad
Because of the great potential of hyperspectral remote sensing in detecting the attributes of ground objects, imaging spectroscopy technology has been paid more and more attention.
(1) airborne hyperspectral imager
1983, the aerial imaging spectrometer (AIS- 1) developed by the United States acquired the first hyperspectral image, which marked the appearance of the first generation of hyperspectral imager. From 65438 to 0987, the Jet Propulsion Laboratory (JPL) of NASA successfully developed the airborne visible/infrared imaging spectrometer (AVIRIS), which marked the emergence of the second generation hyperspectral imager.
(2) Spaceborne hyperspectral imager
In the aerospace field, the Moderate Resolution Imaging Spectrometer (MODIS) in the Earth Observation Program, developed by the Jet Propulsion Laboratory of the United States, became the first on-orbit spaceborne imaging spectrometer with the launch of the TER2RA satellite, and began to transmit images to the ground in 2000.
In 2000, the Hyperspectral Imager (Hyperion) with a ground resolution of 30m carried by the EO2 1 satellite launched by NASA achieved good application results in quantitative mapping of minerals. In 2002, the coastal ocean imaging spectrometer (COIS) carried by NEMO satellite in the United States had the ability of adaptive signal recognition, which met the different needs of military and civilian. In addition, the hyperspectral imaging sensor delivered to Cortland Air Force Base in June 2007 will be loaded into space via Tac2Sat23 satellite.
At present, many countries are actively developing their own hyperspectral sensors, such as EnMAP of Germany's Environmental Monitoring and Analysis Program, MSMI of South Africa's multi-sensor small satellite imager and HERO of Canada's hyperspectral environment and resources observer.
1.2. 1.2 research status of hyperspectral image analysis technology abroad
With the rapid development of imaging spectrometer, the research on spectral database of ground objects and hyperspectral image analysis technology has also developed rapidly.
In terms of ground object spectral database technology, the United States is the most advanced, and the representative ones are JPL standard spectral database, USGS spectral database, ASTER spectral database and IGCP2264 spectral database. In addition, the US Air Force and the Environmental Protection Agency established the AEDC/ Environmental Protection Agency spectral database for the diagnosis of air pollution and air composition, and established the forest hyperspectral database for the HYDICE imaging spectrometer developed by the US Naval Research Laboratory. Other countries have also started the research and construction of spectral database technology. For example, in the early 1990s, Britain established a seawater spectral database for studying seawater color.
The National Aeronautics and Space Administration (NASA), the European Space Agency (ESA), the National Space Development Agency (NASDA) of Japan, universities and research institutes all have specialized research institutions for hyperspectral image application analysis.
Foreign commercial remote sensing image processing systems have successively added imaging spectral data processing modules, among which ENVI of RSI company, PCI of PCI Geomatics company and TNTmips of MicroImages company are the representatives.
1.2. 1.3 Development Status of Hyperspectral Remote Sensing Technology in China
China started to develop its own hyperspectral imaging system in the middle and late 1980s, following the development of international hyperspectral remote sensing technology and combining with the increasing application demand in China. The main imaging spectrometers are push-broom imaging spectrometers (PHI) series, practical modular imaging spectrometers (OMIS) series, high-resolution imaging spectrometers (C2HRIS) developed by Changchun Institute of Optics and Mechanics, Chinese Academy of Sciences, and stable large-field polarization interference imaging spectrometers (SLPIIS) developed by Xi 'an Institute of Optics and Mechanics. The Medium Resolution Imaging Spectrometer (CMODIS) developed by Shanghai Institute of Technical Physics, Chinese Academy of Sciences was launched with Shenzhou III in 2002, and successfully acquired hyperspectral images in space. It obtained 30 bands from visible light to near infrared, and 4 bands from middle infrared to far infrared, with a spatial resolution of 500 meters.
The Chang 'e 1 satellite was launched on June 5438+1October 2007, carrying an interference imaging spectrometer developed by Xi 'an Institute of Optics and Fine Mechanics of China Academy of Sciences. It is used to obtain two-dimensional multi-spectral sequence images of the lunar surface and resolvable earth spectral images. By cooperating with other instruments, the content and distribution of useful elements and substance types on the lunar surface are analyzed, and the distribution map of each element on the lunar surface is compiled by using the obtained data.
From 2007 to 20 10, China will establish a small satellite constellation for environmental and disaster monitoring and prediction, which will carry a hyperspectral imager with an average spectral resolution of 5 nanometers and a ground resolution of 100 meters.
While actively developing imaging spectrometers with independent intellectual property rights, China has also made a series of gratifying achievements in the research of spectral data technology and hyperspectral image analysis technology of ground objects.
In the early 1990s, Anhui Institute of Optics and Mechanics, China Academy of Sciences, Institute of Remote Sensing and other units collected a large number of typical ground objects and established China's first comprehensive "ground object spectral feature database". During the period of 1998, China Land and Resources Airborne Geophysical Prospecting and Remote Sensing Center established the "Typical Rock and Mineral Spectrum Database", which included more than 500 major typical rock and mineral resources in China. In 2000, the Institute of Remote Sensing of Chinese Academy of Sciences developed a typical ground object spectrum database and its management system based on GIS and network technology, recorded more than 10000 ground object spectra, dynamically generated corresponding spectral curves and simulated bands of remote sensors, and realized the link between the spectral database and "3 S" technology.
1.2.2 Introduction of Hyperspectral Imager
1.2.2. 1 Introduction of foreign hyperspectral imager system
(1) aerial hyperspectral imager
1983, the world's first imaging spectrometer AIS-1(Aeroimaging spectrometer-1) was successfully developed in the American Jet Propulsion Laboratory, and successfully applied to vegetation research, mineral mapping and other aspects, showing the world the potential of hyperspectral imaging technology. Since then, the American airborne advanced visible infrared imaging spectrometer (AVIRIS), the Canadian fluorescence line imaging spectrometer (FLI) and the small airborne imaging spectrometer (AIS) developed on this basis, the American Deadalus company's MIVIS, the GER company's 79-band airborne imaging spectrometer (OMI- 10 and OMI-20), and the US Naval Research Institute's laboratory's hyperspectral digital image acquisition tester (HY
Table 1. 1 Information of Main Airborne Hyperspectral Imagers Abroad
In recent years, the successful application of imaging spectroscopy technology in resource investigation, crop growth, plant diseases and insect pests, soil conditions, geological exploration and other aspects has made all countries in the world see the great prospect and potential of this new technology, and some countries with conditions in the world are competing to invest in the development and application of imaging spectroscopy technology. At the same time, many countries refer to the advanced technology of the existing imaging spectrometer, which makes the newly developed system inherit the advantages of the old system, and at the same time, it has been further improved in many aspects, and has made great progress in stability, detection efficiency and comprehensive performance. Among them, the representative ones are Probe in the United States, HyMap in Australia and TEEMS system specially developed by GER Company in the United States for Texaco Oil Company.
Probe- 1 and Probe-2 are another influential aerial imaging spectrometer system developed by earth search science company. The system has 128 bands in the range of 0.4 ~ 2.5 microns, and the spectral resolution is 18 nm.
HyMap is the abbreviation of hyperspectral mapper, which was mainly developed by Australian Intergrated Spectronics Company. HyMap has 126 bands in the spectral range of 0.25 ~ 0.45 μ m, and two optional bands are set in the wavelength regions of 3 ~ 5 μ m and 8 ~10 μ m. * * There are 128 bands. Its data has achieved high performance in spectral calibration, radiometric calibration and signal-to-noise ratio, and the overall spectral calibration accuracy is better than 0.5nm；. . The signal-to-noise ratio of short-wave infrared band (2.0 ~ 2.5 microns) is higher than 500∶ 1, and the signal-to-noise ratio of some bands is even as high as 1000∶ 1.
TEEMS is the abbreviation of Texaco multi-spectral imaging spectrometer for energy and environment. This is a practical hyperspectral imager with more than 200 bands and advanced performance, which was specially developed by American Geophysical and Environmental Research Corporation (ger) at the request of Texaco Technology Company and in cooperation with Texaco experts. The system has the imaging ability of ultraviolet, visible light, near infrared, short-wave infrared and thermal infrared bands, so it has great potential in petroleum geological exploration, especially in exploring the characteristics related to oil and gas reservoirs.
In recent years, the thermal infrared imaging spectrometer has made substantial progress. The most representative is the Space Enhanced Broadband Array Spectroscopy System (SEBASS) developed by American Aerospace Corporation. SEBASS has two spectral regions: middle infrared, 3.0 ~ 5.5 microns, and bandwidth of 0.025 microns; Long-wave infrared, 7.8 ~ 13.5μ m, with a bandwidth of 0.04μm m. There are 100 band and 142 band in the medium-wave infrared region and the long-wave infrared region respectively. The detectors used are two Si: As focal plane 128* 128, effective frame rate 120Hz, temperature sensitivity of 0.05℃, and signal-to-noise ratio > 2000. Thermal infrared imaging spectrometer provides valuable data to better reflect the properties of ground objects, and has been applied in many fields such as prospecting, geological mapping, environmental monitoring, mapping of agricultural and forestry resources, vegetation growth and so on.
(2) Space hyperspectral imager
The United States has successively developed the Moderate Resolution Imaging Spectrometer (MODIS) and EO- 1 hyperspectral satellite, and cooperated with Japan to develop the advanced satellite thermal emission/reflection radiometer and the US military's "Power-Satellite" hyperspectral satellite, which has been far ahead in the research of space imaging spectroscopy technology in the world.
MODIS is the main detection instrument on EOS-AM/KOOC-0/satellite (launched in February, 1999) and EOS-PM/KOOC-0/(launched in May, 2002)-medium resolution imaging spectrometer, and it is also the only live earth observation instrument on EOS Terra platform. The hyperspectral data of 36 bands in the range of 0.4 ~ 14μ m can be obtained by MODIS, which provides an important data source for the comprehensive study of natural disasters, ecological environment monitoring, global environment and climate change and global change.
MODIS is an important sensor on terra and aqua satellites. It is the only on-board instrument that broadcasts real-time observation data directly to the whole world through X-band, and can receive data and use it for free. MODIS can obtain hyperspectral data of 36 bands in the range of 0.4 ~ 14μ m, which provides an important data source for ecological environment research, natural disaster monitoring, global environment and climate change research.
The ASTER space-borne thermal radiation and anti-radiation instrument carried by Terra satellite was launched on 1999 12 18, and was manufactured by the Ministry of International Trade and Industry of Japan. A Japanese-American technical cooperation team is responsible for the calibration confirmation and data processing of instruments. ASTER is the only sensor for high-resolution analysis of surface images. Its main task is to obtain the high-resolution analysis image data of the whole surface-black and white stereo photo through 14 channel. ASTER can image the same area within 4 to 16 days, and has the ability to repeatedly cover the changing areas of the earth's surface. One feature of ASTER data is observation based on user requirements, that is, images can be obtained anytime and anywhere according to user requirements. ASTER's broad spectral coverage and high resolution provide scientists with identification information of many disciplines, such as monitoring the advance and retreat of glaciers, monitoring potential active volcanoes, identifying crops, monitoring cloud morphology and physical conditions, wetland assessment, thermal pollution monitoring, coral reef degradation, surface temperature mapping of soil and geology, and measuring surface heat balance.
NASA's Earth Orbit-1 (EO- 1) is a part of NASA's new millennium plan. June 2000165438+1October 2 1 launched. Earth observation satellite 1 and LandSat-7 cover the same ground orbit, and the detection time of the two satellites on the same ground is about 1 minute. EO- 1 has three basic remote sensing systems, namely advanced land imager (ALI), hyperspectral imager (HYPERION) and linear metal imaging spectrometer array atmospheric correction (LAC). Hyperion, a hyperspectral remote sensor mounted on EO- 1, is the representative of the new generation of space imaging spectrometer, the only spaceborne hyperspectral imaging spectrometer and the only hyperspectral measuring instrument that can obtain data publicly. * * * has 242 bands, the spectral range is 400-2500 nm, the spectral resolution reaches 10nm, and the spatial resolution is 30m.
In July, 2000, the Fourier Transform Hyperspectral Imager (FTHSI) on MightSat-II satellite launched by the United States is a successful example of interferometric imaging spectrometer.
On 200 1 based on the airborne autonomous small satellite PROBA, the European space agency successfully developed a compact high-resolution imaging spectrometer (CHRIS) and successfully launched it. CHRIS has five imaging modes in the imaging range of 4 15 ~ 1050μ m, and the number of bands, spectral resolution and spatial resolution are different in different modes, the number of bands are 18, 37 and 62 respectively, the spectral resolution is 5 ~ 15nm, and the spatial resolution is/kloc-. Chris can observe the ground objects from five different angles (observation modes). This design enables him to obtain the directional characteristics of the reflection of the ground objects.
Following AM- 1 MODIS of the United States, the European Space Agency successfully launched Envisat satellite in March 2002, which is an advanced polar-orbiting Earth observation satellite with a combined large platform. Among them, the resolution imaging spectrometer (MERIS) is a push-broom medium resolution imaging spectrometer with a field of view of 68.5. Its ground resolution is 300 meters, and there are 15 bands in the visible-near infrared spectrum. The layout of spectral bands can be selected and changed by program control.
After ADEOS- 1, Japan launched its successor satellite ADEOS-2 in February 2002, carrying two remote sensors (AMSR and GLI) from NASDA and three remote sensors (Polar, ILAS-2 and Haifeng) provided by international or domestic partners. GLI has 23 and 6 bands in visible-near infrared and short-wave infrared respectively, and 7 bands in mid-infrared and thermal infrared. See table 1.2 for the representative spaceborne imaging spectrometers that have been launched so far.
Table 1.2 Main foreign spaceborne hyperspectral imagers
1.2.2.2 introduction of China hyperspectral imager system
(1) aerial hyperspectral imager
The development of imaging spectrometer in China has gone through the development process from multi-band scanner to imaging spectrum scanning, from electromechanical scanning to solid-state scanning of area CCD detector.
During the Eighth Five-Year Plan period, a new modular aerial imaging spectrometer (MAIS) was successfully developed, which marked a major breakthrough in the technology and application of aerial imaging spectrometer in China. Since then, push-broom imaging spectrometer (PHI) and practical modular imaging spectrometer system (OMIS) developed by China have occupied an important position in the world aviation imaging spectrometer family.
(2) Space hyperspectral imager
China launched the Shenzhou-3 unmanned spacecraft in March 2002, carrying a Moderate Resolution Imaging Spectrometer (CMODIS), which has 34 bands with the wavelength range of 0.4 ~12.5μ m. In addition, the environmental disaster reduction satellite is equipped with a hyperspectral remote sensor with the band of 1 15. The middle resolution imaging spectrometer carried by FY-3 meteorological satellite has 20 bands, and the imaging range includes visible light, near infrared, middle infrared and thermal infrared. Chang 'e-1 satellite is equipped with an interferometric imaging spectrometer developed by China to detect lunar materials.
1.2.3 hyperspectral remote sensing imaging characteristics and data expression
The images obtained by hyperspectral imaging contain abundant triple information of space, radiation and spectrum. Its main feature is to combine the traditional image dimension and spectral dimension information into one, and obtain the continuous spectral information of each ground object while obtaining the surface spatial image. Hyperspectral data is a cube of spectral image, which consists of spatial image dimension, spectral dimension (a "continuous" spectral curve can be obtained from each pixel of hyperspectral image) and feature space dimension (hyperspectral image provides a hyper-dimensional feature space, and mining hyperspectral information requires a deep understanding of the distribution characteristics and behavior of ground objects in the N-dimensional feature space formed by hyperspectral data).
1.2.4 Main application fields of hyperspectral remote sensing
Because hyperspectral remote sensing can provide more detailed spectral information, some scholars have extended the research of hyperspectral remote sensing from the initial mineral identification to the exploration of water, vegetation and ecology, environmental resources and so on. But at present, it is mainly concentrated in the research fields of geology, vegetation and water environment.
Application of 1.2.4. 1 in vegetation monitoring
Hyperspectral remote sensing provides strong support for vegetation parameter estimation and analysis, vegetation growth monitoring and yield estimation with its ultra-high spectral resolution.
1) The "red edge" effect of plants: "REP" is the point where the spectral curve of green plants changes fastest between 680 and 740 nm, and it is also the inflection point of the first derivative spectrum in this interval. "Red edge" is the most typical feature of plant spectral curve, which can well describe the health and pigment state of plants. When the "red edge" moves in the infrared direction, it can generally be judged that the green plants have high chlorophyll content and vigorous growth vitality; On the contrary, when the "red edge" moves in the direction of blue light, it may generally be that plants are in an unhealthy state such as yellow leaves due to lack of water and other reasons. When the plant coverage increases, the slope of "red edge" will become steeper.
2) Vegetation index: Vegetation index mainly reflects the difference between the reflection of vegetation in visible light and near infrared band and soil background, and each vegetation index can be used to quantitatively explain the growth of vegetation under certain conditions, which is an effective method to monitor the growth and distribution of ground plants and evaluate vegetation qualitatively and quantitatively by using remote sensing spectral data. According to different research purposes, people put forward dozens of vegetation indices, such as specific vegetation index RVI, normalized vegetation index NDVI, differential environmental vegetation index DVIEVI, vertical vegetation index PVI, soil-adjusted vegetation index SAVI and so on.
Application of 1.2.4.2 in agriculture
The application of hyperspectral remote sensing in agriculture is mainly manifested in rapid and accurate extraction of crop growth information, crop growth monitoring, estimation of primary productivity and biomass of vegetation (crops), estimation of light energy utilization rate and evapotranspiration, remote sensing monitoring and prediction of crop quality, so as to adjust the input of materials accordingly and achieve the purposes of reducing waste, increasing yield, improving quality and protecting agricultural resources and environmental quality. There are two main methods to estimate agricultural parameters of crops by hyperspectral remote sensing data: one is to establish the relationship between spectral data or vegetation index derived from it and agricultural parameters of crops by multiple regression method; The second is to estimate the phenological characteristics and agronomic parameters of crops through the red edge parameters of crops.
Application of 1.2.4.3 in atmosphere and environment
With its ultra-high spectral resolution, hyperspectral remote sensing can identify spectral differences caused by changes in atmospheric composition that broadband remote sensing cannot identify, making it possible for people to quantitatively analyze the surrounding ecological environment by hyperspectral remote sensing. Hyperspectral technology can be used to detect the difference of chemical substances in polluted areas, so as to determine the polluted areas and pollution reasons; Hyperspectral images can also be used to detect harmful environmental factors, such as accurately identifying harmful waste minerals, drawing the distribution map of special altered minerals, evaluating the risk level of wildfire, and identifying and detecting burning areas.
Application of 1.2.4.4 in Geology
Geological and mineral survey is the most successful field of hyperspectral remote sensing application. The diagnostic spectral characteristics of various minerals and rocks on the electromagnetic spectrum can help people identify different mineral components. In geology, it is mainly used to detect the diagnostic characteristics of absorption and reflection of rocks and minerals, so as to classify, map and explore rocks and minerals.
Military application of 1.2.4.5
Hyperspectral images contain abundant information of ground objects, which can be used to accurately identify the types of ground objects, and have been successfully applied to military reconnaissance, identification and camouflage. The hyperspectral imager designed by the US Navy can provide 2 10 imaging spectral data in the spectral range of 0.4 micron to 2.5 micron, and can obtain the dynamic characteristics of offshore environmental targets, such as seawater transparency, ocean depth, ocean atmospheric visibility, ocean currents, tides, seabed types, bioluminescence, beach characteristics, underwater hazards, oil spills, total water vapor in the atmosphere, cirrus clouds, etc., which is very useful for offshore operations.