The development of precision farming urgently requests that remote sensing technique offers to timely and accurate ground information. Soil water content, soil organic matter content, soil roughness and soil texture etc. are very important information in precision farming. As hot point and frontier in remote sensing, hyperspectral remote sensing technique not only has the advantages of traditional remote sensing that can timely and undisturbedly be used to detect large area crop, but also has special advantages. It has very high spectral resolution. More delicate spectral difference of crops can help us to precisely classify crops types and to monitor and analyze crops’ vigor and the environment factors that affect crops’ product. Hyperspectral remote sensing has great potential of quantitatively retrieving for objects’ characteristics.
This thesis focuses on extracting soil information from hyperspectral data, and puts great emphasis on the study of retrieving soil characteristics from laboratory spectra. The first chapter mainly introduced the background of hyperspectral remote sensing and precision farming, and then, introduced the applications and perspectives of hyperspectral remote sensing in precision farming. In the second chapter, we primarily introduced the measurement of soil characteristics and soil spectra in laboratory, and analyzed feature of soil spectra. The third chapter is the most important part of this thesis. We discussed soil spectral properties. It included: 1) The relationship between soil minerals and soil spectral reflectance; 2) The relationship between soil color and soil spectra as well as inversion of soil color from spectral reflectance; 3) The relationship between soil surface moisture and soil spectral reflectance as well as evaluation of several inversion method of soil surface moisture from reflectance; 4) The relationship between soil organic matter and soil spectral reflectance as well as inversion of soil organic matter and soil spectral reflectance; 5) The relationship between soil texture, soil ferric oxide and soil spectral reflectance. The fourth part studied the BRDF properties of soil and with two models inverse models’ parameter of soils. The fifth part introduced the imaging mechanism of remote sensing and the spectra and radiance calibration methods for remote sensing images, as well as inversion of soil characteristics from airborne remote sensing image. The sixth chapter summarized the whole thesis and listed the achievement of this study, as same as, pointed out the difficulties in precise inversion of soil characteristics from hyperspectral image.Main development and conclusion as follows:
(1) By analyzing a large number of soil spectra, we found except at the obvious absorption position, the line of these points’ reflectance at the wavelengths 400, 600, 800, 1350, 1800, 2100 and 2400 nm are fitted well with spectral curve. This is useful for soil spectral data compressing and band selecting.
(2) From the correlation between soil spectral reflectance and soil color, we utilized regression model to forecast soil Munsel properties.
(3) The relationship between normalized soil reflectance and moisture was investigated.
For all the wavelengths and all the soils, results show that for low soil moisture levels, the reflectance decreased when the moisture increased. Conversely, after a critical point, soil reflectance increased with soil moisture. For some soils, the reflectance of the wettest conditions can overpass that of the driest conditions. For both low and high soil moisture levels, and the seven wavelengths selected, the relative reflectance was strongly correlated with moisture. Adjustment of the relationships over individual soil types provides better soil moisture retrieval performances.
(4) The normalization of reflectance approach, derivative approaches and the difference approaches were used to forecast soil surface moisture. And the best overall retrieval performances were achieved with the absorbance derivatives and the difference of absorbance.
(5) By analyzing the relationship between soil organic matter and soil reflectance, we forecasted soil organic matter and verified the performances of models.
(6) By using BRDF model we analyzed the BRDF property of different soil at different moisture.
(7) Retrieved soil characters from hyperspectral image and developed the soil characteristics map for precision farming.
From the beginning of remote sensing, imaging technology has advanced in two major ways: one is the improvement in the spatial resolution of images; another is the improvement in the spectral resolution of images. Conventional multispectral scanners record up to 10 or so, spectral bands with bandwidths on the order of 0.10μm in visible to short wave infrared bands. Furthermore, hyperspectral imaging, or called imaging spectrometry, can acquire images in hundreds of registered, contiguous spectral bands such that for each picture element it is possible to derive a complete reflectance spectrum.
Hyperspectral remote sensing effectively make the spectral feature and geometric characters of objects together. From the view of earth observation from space, hyperspectral data provide human being more abundant information, not only in the deep explorations of object’s physical and chemical characters, but also in the precise classification of different objects and knowledge innovation. In case of so much spectral bands and such huge quantities of data, some conventional data processing methods cannot play good roles. Aiming at the hyperspectral image cube, the understanding and data processing in image spatial dimension must be changed to that completed in the spectral dimension.
This dissertation is just concentrated on above aspects and evolved in the systematic and innovative views. This dissertation begins from the introduction on hyperspectral remote sensing technology. In the second and third chapters, two key points in hyperapectral data processing and analysis area, hyperspectral data calibration and parameterizationand, and hyperspectral image classification and identification, were dissertated. The fourth and fifth chapter pays more attentions to the hyperspectral data mining supported by the temporal and spatial information. In general, this study has some advantages as follows:
(1) As for spectral feature selection, spectral bands selection and objects quickly finding in image cube were provided. On the other hand of spectral featureextraction, several selections of spectral parameterization were also provided. Considering the hyperspectral geological remote sensing, stratum spectral histogram was established specially for 14 strata in Tulufan anticline.
(2) After discussion on the traditional image classification, a new method, Expert Decision Classification Based on Feature Optimization, was provided here. It is designed out in accord with two principles: one is the spectral feature optimization and parameterization, another is fuzzy and expert decision in pixel identification. Comparing with other method, this method can acquire more accurate classification results.
(3) Several spectra of man-made camouflage materials were provided here. In the SWIR, the position and relative intensities of the major absorption features associated with water are difficult to duplicate due to the complex architecture of vegetation. In addition, convex geometry projection was successfully used in the different metal material detection.
(4) On the bases of vegetation spectral analysis and hyperspectral vegetation indices, Multi-temporal Indices Image Cube was put forward and used in the dynamic growing analysis of Japanese lettuce, Chinese cabbage, and wheat stressed by nitrogen or water contents.
(5) In the area of hyperspectral data analysis supported by spatial information, Four application aspects were provided: spatial fusion based spectral reversion, hyperspectral data analysis associated with pixel position analysis, spectral unmixing and classification in the field patch units, and image classification supported by digital geomorphology model.
The data of the hyperspectral remote sensing has the characteristic of combining the image and spectrum, which is the powerful supplement for the traditional data resource such as aerial picture and multi-spectral image. We study on the image information and quantitative analysis focusing on the spectral dimension because the hyperspectral data has the abundant spectral information. During transacting the hyperspectral image, the spectral mapping analysis is one of the key techniques to identify and classify the material in the image. In this paper, I aim to build a useful and effective database fitting with the hyperspectral data, then, on the base of this database, contrast and value some spectral mapping technique in common use, and then develop and validate a new spectral mapping technique. So at first this paper introduces the existing hyperspectral database briefly and its important functions, then with the standard spectrums in the existing database, do some study on the value of some typical spectral mapping technique and discuss the results of the different techniques, at last, noticed that in the hyperspectral remote sensing, the continuum removed method is very useful but used only with the spectrum of a single pixel to analyze spectrum and extract the feature bands useful with the classification, so in this paper, based on the continuum removed arithmetic, I programmed with VC++ to fulfil the functions of the continuum removed to the whole hyperspectral image, normalization and extracting the feature space for the classification. Then, aiming at the former image and the after-continuum removed image, the classification results of the MLC and SAM are compared. The results showed that the new technique is effective to identify the material with the spectrum which has the distinct absorbing and reflecting characteristic, but it’s hard to use with the material having no characteristic of the spectrum. So this method needs to be improved in the future.
The major objective this thesis is the key technique research of the airborne multispectral digital camera system and its application on the remote sensing of oriental migratory locust. The main research works are list bellow：
1. An airborne multispectral digital camera system has been successfully developed based on area CCD sensors and other peripheral facilities. In fact, the whole system includes three different multispectral camera systems, Basler cameras with optical filter system, MS4100 high-resolution 3-CCD digital multispectral cameral and Kodak professional DCS760 camera. Those systems have their independent hardware and control software, and they can be used either in combine mode or separate mode. IEEE-1394 Bus and Camera Link technique are used to capture the image data. IEEE-1394 Bus and Camera Link are the latest advance in communications interface technology for transmitting digital data. The computer, controlled by a executable program under windows operating system, provides all control functions including exposure time control, exposure interval control, retrieving images from grabber, saving images to disk array, displaying captured images for real time monitor and recording GPS information. The GPS receiver provide 1PPS pulse signal to trigger circuit, and provide the GPS locations to the control computer as well. The control programs are developed using VC++6.0, and compatible with WINDOWS NT/2000 operating system.
2. For the first time in China, the airborne multispectral digital camera system was used to research the remote sensing of the oriental migratory locust. After analyzing the biological and environmental characteristics of the locust, the detect method was proposed that we can monitor the spectral changes of the vegetable caused by the locusts. In august 2002, a flight experiment had been done to monitor the locust area in Nandagang farmland in Hebei province. A large amount of excellent image were acquired in this experiment. And the result denoted that the airborne remote sensing method was feasible. On the other hand, the result indicated that the newly designed digital camera systems were successful.
3. For frame transfer CCD sensor, in the phase of the frame data transferring, the light source still hit the sensors surface while part of the image was moving toward the light shielded region. So the image was polluted, and we call this frame transfer smear. In this thesis, a restoration algorithm has been proposed to recover the frame transfer smear. And effect of this algorithm had been testify by many ground and flight images.
4. On the radiometric correct process of multispectral image, an in-flight statistic template correcting method was introduced. And this method was testified using 1431 multispectral images acquired in the test flight.
5. After analyzing the difficulties and disadvantages of the optical band register, image register method was adopted instead of the optical register. This is implemented by using the precise hardware synchronic trigger control circuit and the after flight image process. As a result, the camera heads do not need to be aligned at the precision of pixel level. The only requirement is that the field of view of all cameras is roughly the same. One more advantage of image register is that the replacement of different lens and optical filters would not be limited in this system. This will greatly enhance the system performance and adjustability.