Acquisition of remote sensing spectral data

Remote sensing technology is developed from aerial photogrammetry and has gone through three stages:

1. Development stage of aerial photogrammetry

The earliest aerial photograph still preserved is the Boston photograph taken by J.W. Blake from a balloon in 1860. The application in geology began at 19 13. At that time, someone photographed Benglasen oil field in Libya, Africa, and compiled the geological map of Benglasen oil field with this group of dirty photos. Aerial remote sensing mainly takes airplanes or balloons as vehicles, uses aerial cameras to obtain the information of targets, and then obtains the final aerial photos through positive and negative processes. Aerial photography uses the panchromatic band of electromagnetic wave and visible light, and uses photographic film to receive the sunlight reflected by the photographed object for photosensitive imaging. Generally speaking, the photosensitive range of photographic film is 0.3 ~ 0.9 micron ... In most cases, aerial photography is vertical photography, that is, the main shaft of the aerial camera keeps taking pictures in the vertical direction; Under special circumstances, oblique photography is carried out with a special camera. Aerial photography can be divided into four types according to the characteristics of electromagnetic wave band, corresponding photosensitive film and image, namely: full-color black-and-white image of aviation visible light; Aerial visible light true color image: aerial infrared false color image: aerial infrared black and white image. Among them, the most commonly used are aerial visible full-color black-and-white images and aerial infrared false-color images, which mainly use the broadband reflection intensity characteristics of ground object spectra.

2. Multispectral satellite remote sensing station

Digital satellite imaging begins with meteorological satellites. In 1960, TIROS- 1 meteorological satellite provided very rough satellite images, which were mainly used to show cloud types. Subsequently, in the1970s, the National Oceanic and Atmospheric Administration (NOAA) of the United States launched a very high resolution radiation sensor (AVHRR) for weather forecasting, with a ground resolution of 1. 1km. We saw the cloud picture it got in the TV weather forecast program. At the same time, starting from 1970, some satellites with higher resolution sensors have been launched one after another. For example, 1972 On July 23rd, the National Aeronautics and Space Administration (NASA) launched the first Earth Resources Technology Satellite (ERTS-U) dedicated to monitoring and mapping the earth's surface, and renamed it Landsat on 1975. Landsat 1-3 is equipped with multi-spectral scanner (MSS), which has four bands, namely green, red and two infrared bands, and the ground resolution is about 80m. 1982, Landsat4 was equipped with thematic imager (TM), which has seven bands, wider than MSS, and the band width is divided more finely, which can better reflect the changing law of spectral characteristics of ground objects. Its ground resolution is 30m except for the 6th band, which is 120m. The most typical feature of multispectral remote sensing is that multiple spectral features of the same target can be obtained simultaneously by using multiple bands. This greatly improves the ability of remote sensing to identify ground objects. Subsequently, countries have followed suit, and the spectral range of the sensor ranges from visible light, infrared to microwave band, and its application range is also expanding.

3. Development stage of imaging spectral remote sensing technology

Imaging spectral remote sensing technology is a leap in the development of multispectral technology. Hunt's research results show that the absorption width of characteristic minerals is about 20 ~ 40 nm, while the spectral resolution of multi-spectral remote sensing data (such as MSS and TM) is only about 100nm, so remote sensing scientists begin to study remote sensing sensors with high spectral resolution and spatial resolution. 198 1 year, a space shuttle multispectral infrared radiometer (SMIRR) observed the earth's surface with the American space shuttle Columbia, which realized the direct identification of carbonate rocks and clay kaolin minerals from space by hyperspectral remote sensing for the first time, thus opening a new chapter in lithologic identification by imaging spectral remote sensing. Following the successful development of AIS- 1 and AIS-2 of JPL and AVIRIS aerial imaging spectrometers, several imaging spectrometers such as FIL/PML, CAS 1 and SFSI (Tong Qingxi et al., 1993) have been successfully developed. Others are: HIRIS (High Resolution Imaging Spectrometer), which has 192 spectral bands in the range of 0.4 ~ 2.5 microns, with a ground resolution of 30m and a spectral resolution of 9.4nm in the wavelength range of 0.4 ~ 1.0 microns. In the range of 1.0 ~ 2.5 microns, there are168 ... kerekes & landgrebe, 199 1). The 63-channel imaging spectrometer (GER) of Geophysical and Environmental Metals Research Company is specially designed for geological remote sensing research and has been used for lithologic mapping many times. Bobby W. Rockwell, 1997). In addition to aerial imaging spectrometers, both the United States and the European Space Agency (ESA) have made plans to develop space imaging spectrometers, among which the American Moderate Resolution Imaging Spectrometer (MODIS) has been launched into orbit with the Earth Observation System (EOS) to realize periodic remote sensing observation of the earth with high spectral resolution. The Medium Resolution Imaging Spectrometer (MERIS) of ESA will also be launched at the same time (Tong Qingxi et al., 1993).

From 1990 to 1995, Roger N.Clark and others used AVIRIS data to identify and draw minerals and lithology at Capulet proving ground in Nevada, USA. They found that the imaging spectrometer can not only distinguish the overall brightness and slope difference in the surface emission spectrum (the basis of multi-spectral technologies MSS, TM and SPOT to distinguish ground objects). Moreover, the spectral absorption band used to identify special ground objects can be obtained, and the spectral analysis of imaging spectral data can identify and map any substance (minerals, vegetation, human T-object, water, snow, etc. ) has unique absorption characteristics in the measured spectral range (Clark, R.N. et al., 1996).

Shanghai Institute of Technical Physics, Chinese Academy of Sciences is the main research institution of imaging spectrometer in China. 1983, the first six-channel infrared subdivision spectrum scanner working in the short-wave infrared spectrum region (2.05 ~ 2.5 microns) was successfully developed, and its spectral resolution was between 30 ~ 50 nm. 1987, driven by the national and Chinese Academy of Sciences gold prospecting task, the instrument developed to 12, and its band position tends to be consistent with the absorption band of surface clay minerals and carbonate minerals, so it has greater ability in geological lithology identification (Tong Qingxi et al., 1993). There are also thermal infrared multispectral scanner (TIMS), 19 band multispectral scanner (AMSS) and 7 1 band multispectral airborne imaging spectrometer (MATS). The data of these spectrometers are mainly used for remote sensing of oil and gas resources (Zhu Zhenhai, 1993) and mineral mapping (Wang Jinnian et al., 1996), and the data processing technology and theoretical research of mineral identification have made different progress (Li, 1997).

Looking at the acquisition of remote sensing spectral data, there are several new developments:

(1) expands the application spectrum range and increases the spectral band; (2) spectral and spatial resolution are improved; ③ It has the function of obtaining stereo image pairs, which breaks the ability that only aerial photos can have stereo image pairs (such as SPOT images); (4) improving detector performance or detector device, namely linear array and area array CCD device; ⑤ The accuracy of image data is improved; ⑥ Vertical expansion of application fields, such as direct identification of hematite, goethite and other minerals using TM image data.

At the end of the 20th century and the beginning of the 20th century, the space hyperspectral imaging satellite has become an important frontier technology of remote sensing and earth observation, playing an increasingly important role in studying the earth's resources and monitoring the earth's environment.

The development of hyperspectral resolution remote sensing technology is one of the major technological breakthroughs made by mankind in the last two years of the 20th century 10. It is the frontier technology of remote sensing at present and even at the beginning of 2 10 century. The ground image obtained by hyperspectral imaging contains rich spatial, radiation and spectral information. In the late 1990s, with the solution of a series of basic problems in the application of hyperspectral remote sensing, such as the calibration and quantification of hyperspectral imaging information, the visualization and multi-dimensional expression of imaging spectral image information, image-spectrum conversion, and the processing of large amount of data, hyperspectral remote sensing gradually shifted from the experimental research stage to the practical application stage. As a hot spot in the application of hyperspectral remote sensing, the improvement of hyperspectral data information mining technology and the expansion of its closely related application fields are the key points.

The most important feature of hyperspectral remote sensing data is that it combines the traditional image dimension and spectral dimension information, and obtains the continuous spectral information of each ground object while obtaining the surface spatial image, thus realizing the inversion of ground object component information and ground object recognition according to the spectral characteristics of ground objects. It consists of the following three parts:

(1) spatial image dimension

In the dimension of spatial image, hyperspectral data is similar to general images. The general pattern recognition algorithm of remote sensing image is a suitable information mining technology.

(2) Spectral dimension

A "continuous" spectral curve can be obtained from each pixel of hyperspectral image, and the "spectral matching" technology based on spectral database can realize the purpose of identifying ground objects. At the same time, most ground objects have typical spectral waveform characteristics, especially the spectral absorption characteristics are closely related to the chemical composition of ground objects. The extraction of spectral absorption characteristic parameters (absorption wavelength position, absorption depth and absorption width) will become the main aspect of hyperspectral information mining.

(3) Feature space dimension

Hyperspectral images provide a high-dimensional feature space. For hyperspectral information mining, it is necessary to deeply understand the distribution characteristics and behavior of ground objects in the two-dimensional feature space formed by hyperspectral data. It is found that the high-dimensional space of hyperspectral data is quite empty, the data distribution is uneven, and it tends to concentrate in the corner of the hyper-dimensional cube space. Differences in typical data can be mapped to a series of low-dimensional subspaces. Therefore, it is urgent to develop effective feature extraction algorithms to find low-dimensional subspaces that maintain important differences, so as to effectively realize information.