Transformation of the Hyperspectral Line Scanner into a Strip Imaging System

Transcription

1 Transformation of the Hyperspectral Line Scanner into a Strip Imaging System S. Semanjski Faculty of Transport and Traffic Sciences, University of Zagreb, Croatia, silvio.semanjski@gmail.com D. Gajski Faculty of Geodesy, University of Zagreb, Croatia, dubravko.gajski@zg.t-com.hr M. Bajic Faculty of Geodesy, University of Zagreb, Croatia, milan.bajic@zg.t-com.hr Keywords: hyperspectral, line scanner, parametric geocoding, reflectance, IMU/GPS ABSTRACT: Under European Commission supported scientific project: Airborne minefield area reduction (ARC), that lasted from year 2001 to 2003, several electro-optical digital sensors and the acquisition systems were obtained and the acquisition software Recorder was developed. Among the acquired sensors, a hyperspectral line scanner ImSpector V9 (for the wavelengths from 430 nm to 900 nm), along with an insolation collecting unit (Fodis), was also purchased and lately used in this project. The scanner was used for the acquisition of the reflectivity samples of the mine suspected areas over few different types of terrain whereas the quality of data was limited by several factors. This limitation was the reason to evolve the characteristics of the airborne hyperspectral remote sensing system, by use of V9 in its core, within the frame of the technologic project System for the Multisensor Airborne Reconnaissance and Surveillance in the Crisis Situations and the Environment Protection, in accordance with the foreseen applications. The following kinds of applications are foreseeable: a) measuring the radiance at discrete samples (static or in a direction of flight), b) measuring the reflectance at the discrete samples (static or in a direction of flight), c) imaging the radiance of the area in a form of the strip across the flight direction, d) imaging the reflectance of the area in a form of the strip across the flight direction. While during previous use ( ), when only GPS data (1 Hz to up to 5Hz) were available, thus without capability of measuring the sensor s orientation parameters, in a novel solution a new Position and Orientation System has been applied and combined with the parametric geocoding system program (ParGe). Mentioned achievement in advancing the features of the airborne hyperspectral remote system, enabled wider kinds of the hyperspectral imaging applications. 1 INTRODUCTION The measurement and the acquisition of the hyperspectral data started in Croatia in 2002., [1], [3], and the simple reflectivity sampling which dominated at the beginning was transformed into an airborne imaging hyperspectral acquisition system on board of the helicopter, [9], [6], [13]. Since the radiance, what the sensor measures directly, depends on illumination intensity, geometry, surface roughness and atmospheric conditions, usefulness of the radiance measurement is limited due to its strong dependence on the illumination which is changing during the acquisition mission. In other hand, the reflectance is the ratio of the amount of the electromagnetic waves leaving/striking the target, and it represents a property of the observed matter which is equivalent under different illumination conditions. The basic measuring properties of V9 are determined by its design. Depend-

2 ing on its configuration, the use of different slit widths narrow slit (at the front end of the optical part of the system), enables different spectral resolutions and number of wavelengths bands in spectral range of 430 nm to 900 nm, to be used. When the scanner is directed at nadir to the ground, the area mapped below the scanner is a narrow strip whose dimensions is the function of height above ground and slit width. The V9 s associated digital camera at the output plane provides different combinations of pixel numbers and radiometric range, which dictates requirements toward the acquisition system. The spatial accuracy of the discrete measurements of the imaging depends on the dynamics of the aerial platform and the accuracy of the Position and Orientation System capable to quantify the measurements of aerial platform dynamics. While during previous use ( ), when only GPS data were available (1 Hz or even several Hz), in a novel solution a new Position and Orientation System has been applied and combined with the parametric geocoding system program (ParGe). This process and outcomes are outlined in the next text. 2 SAMPLING THE RADIANCE AND THE REFLECTANCE The hyperspectral line scanner ImSpector V9, with an insolation collecting unit (Fodis) and the digital acquisition unit were obtained in the frame of the scientific projects supported by the European Commission Airborne minefield area reduction (ARC), (from 2001 to 2003). In the same project was developed the acquisition software Recorder, [5]. The scanner was used for the acquisition of the reflectivity samples of the mine suspected areas in several different types of terrain, [2], on board of the helicopter Bell-206, [1]. The quality of data was limited by several factors, [3]. This was the reason to advance characteristics of the airborne hyperspectral remote sensing, by use of V9, within the frame of the technologic project System for the Multisensor Airborne Reconnaissance and Surveillance in the Crisis Situations and the Environment Protection, in accordance to the foreseen future applications, [9], [6]. There are foreseen following kinds of applications: a) measuring the radiance at discrete samples (static or in a direction of flight), b) measuring the reflectance at the discrete samples (static or in a direction of flight), Fig. 1, c) imaging the radiance of the area in a form of the strip in the flight direction, d) imaging the reflectance of the area in a form of the strip in the flight direction. Radiance is what the sensor measures directly and it depends on illumination intensity, geometry, surface roughness and atmospheric conditions. The usefulness of the radiance is limited due to its strong dependence on the illumination which can change during the acquisition mission. Thus we did not apply a system to measure the radiance. The reflectance is the ratio of the amount of the electromagnetic waves leaving the target to the amount of the electromagnetic waves striking the target. It is a property of the observed material and is equivalent under different illumination conditions. The basic measuring properties of V9 are determined by its design and throughput of the digital acquisition system, [3]. Current configuration uses narrow slit (8.8mm x 50µm) at the front end of the optical system, enables spectral resolution of 4.4nm of nearly 106 channels in spectral range from 430 nm to 900 nm. When the scanner is directed at nadir to the ground, the area mapped below the scanner is a narrow strip that has dimensions 0.333H x H, where H is height above ground. The V9 s associated digital camera at the output plane provides different combinations of pixel numbers and radiometric range (from 640x512 pixels, 8 bits to 1280 x 1024 pixels 12 bits) and dictates requirements toward the acquisition system.

3 Figure 1. Meaurement of the reflectivity, [3]. Figure 2. The data recorded for one spatial line contain the spectral information of the radiance and the insolation, [3]. The Fig. 3 and Fig. 4 are examples of the measuremets before the advancement of the hyperspectral line sampling system into the imaging system in the frame of the technology project [9], [6]. Figure 3. The hyperspectral reflectivity samples of the minefield No in Glinska Poljana [2], acquired by the airborne system [1]. Figure 4. The hyperspectral reflectivity samples (the change of the reflectivity of the mistletoe removed from the fir) [10], acquired by the ground based system.

4 3 THE ADVANCEMENT OF A HYPERSPECTRAL SAMPLING INTO AN IMAGING SYSTEM The measurement of the spectral characteristics of the different kinds of vegetation was performed by the ground based system, whose data agree well with [7]. The calibration of the airborne measurement of the reflectivity can be performed in accordance to [11]. The spatial accuracy of the airborne discrete measurements or those of the imaging depends on the movements of the aerial platform, accuracy of the positioning and orientation system. While in the previous use (years to 2003.) were available only GPS data (1 Hz or even several Hz), an advancement is foreseen and realised in the technology project [9], [6]. The main advancements are: a) application of the parametric geocoding and orthorectification of Airborne Scanner Data [12], b) application of the Positioning and Orientation system [13], use of the spectralon reflectance target for the calibration of the radiometry [8]. The advanced features of the airborne hyperspectral remote system enable wider kinds of the applications. 3.1 Imaging geometry requirements To utilize the line scanner in full imaging mode, means acquiring contiguous scan lines (without gaps between them), it is necessary to find the optimum ground speed of the aircraft, which is function of desired Ground Sampling Distance (GSD) and scanner imaging frequency, according to simple equation (1), used to arrive at optimum distance per second: where GS = Ground Speed of the aircraft in [m/s], GSD = Ground Sampling Distance in [m] and fi = Imaging scan period in [s]. Additional constraint for finding the optimum ground speed is spatial resolution along the aircraft track which is determined by the line width and integration time per one scene line (=one detector frame). Actual configuration of ImSpector V9 line scanner, depending on used vertical binning of CCD operation (x1 or x2) and exposure time (i.e. 50ms), provides the integration time per scene line of ms or 90.32ms, respectively. Thus, predicted spatial resolution along track for Zagreb/Jarun Lake campaign was 1.34m. Above described mapping geometry depends on height of the aircraft above ground level (pixel scale factor) which could be obtained from a Digital Elevation Model (DEM). Finally, high imaging resolution of pushbroom scanners, both spectral and spatial, as in case of ImSpector V9 (i.e. Zagreb/Jarun Lake flight campaign using Bell B206 helicopter had produced, in average, the lines with length of 218m and 1.63m wide, with across track pixel size of 19cm at altitude of 750m; in full spectral range with bandwidth of 80 channels and with spectral resolution of 4.4nm), requires additional technology and methods for final exploitation of their produced data, by using mentioned POS and performing the direct georeferencing. The application required a Position and Orientation System capable to measure and record inertial orientation and GPS position data at high frequency, according to general requirement for line imaging mode utilization of hyperspectral imaging scanner, which states that read-out IMU frequency should be at least of those of scanner imaging frequency. Thus, Position and Orientation System, is built around chosen tactical-grade MEMS imar ivru-rssc, so-called Vertical Reference Unit, with integrated internal GPS unit and strap-down processor. Latest mentioned proved as self-sufficient system capable to deliver at least one set of exterior orientation parameters (Roll, Pitch, Yaw, Latitude, Longitude, Altitude) for each scan line, allowing for direct georeferencing of each separate line to be performed and by time domain transforming these lines into contiguous mapping, or imaging mode (in opposite to the sampling mode performed during earlier use of the System). Transformation of the hyperspectral line scan from sampling to imaging system, along with associated Position and Orientation System, resulted in successfully production of first geocube of Zagreb/Jarun Lake area, (Fig. 5.). (1)

5 Figure 5. One channel of geocoded hypercube of Jarun Lake produced after Zagreb/Jarun Lake flight campaign with ImSpector V9 in full imaging mode (Projection: Transverse Mercator; Datum: Hermannskögel) 4 FURTHER APPLICATION OF THE HYPERSPECTRAL IMAGING SYSTEM Further application of the developed airborne hyperspectral imaging system could be sea oil pollution detection and measurements, [4], forest health research, research of the traffic influence of the motor ways on the environment, water quality monitoring etc. ACKNOWLEDGEMENTS This work was made as a part of the technologic project TP-06/ , System for the multisensor airborne reconnaissance and surveillance in the crisis situations and the protection of environment, supported by the Ministry of science, education and sports, of the Republic of Croatia. REFERENCES [1] Bajić, M., Survey of suspected mined areas from a helicopter, Journal of Mine Action, James Madison University, Issue 7.3, pp [2] Bajić, M., Pračić, Ţ., Vuletić, D., Hyper spectral acquisition in CDA (continuous Data Acquisition), internal technical, ARC IST ,Version 1.0.0, [3] Bajić, M., Gold, H., Pračić, Ţ., Vuletić, D., Airborne sampling of the reflectivity by the hyper spectral line scanner in a visible and near infrared bands, Proceedings of the 24th EARSeL Symposium, New Strategies for European Remote Sensing, Dubrovnik, Croatia, May 2004, Millpress, Rotterdam, 2005, pp [4] Bajić, M., Krajnović, M., Krtalić, A., Airborne hyperspectral detection and geographical mapping of oil slicks in the sea, TIEMS Workshop Croatia 2006, Improvement of Disaster Management Systems - local and global trends, Trogir Croatia, September [5] European Commission, Airborne Minefield Area ReduCtion ( ARC), European Commission Project IST , [6] Fiedler, T., Gold, H., Pavković, N., Milošević, D.I., Gajski, D., Bajić, M., System for the Multisensor Airborne Reconnaissance and Surveillance, II Kongres hrvatskih znanstvenika iz domovine i inozemstva, Split, Svibanj, 2007.

6 [7] Ghassem, A., Theory and applications of optical remote sensing, J. Wiley & Sons, New York. [8] Labsphere, Spectralon Diffuse Reflectance Targets, [9] MZOŠ RH, 2007., System for the Multisensor Airborne Reconnaissance and Surveillance in the Crisis Situations and the Environment Protection, Technology Project TP-06/ of Croatian Ministry of Science, Education and Sports, Faculty of Geodesy of University of Zagreb, HCR Centre for testing, development and training Ltd., Faculty of Transport and Traffic Sciences of University of Zagreb, Pastor Group Zagreb. [10] Pernar, R., Bajić, M., Vuletić, D., Idţojtić, M., Ančić, M., Seletković, A., Aerial high resolution imaging of the mistletoe for the assessment of forest decline in fir stands, Proceedings on CD of the ISPRS Hannover Workshop 2005, High-Resolution Earth Imaging for Geospatial Information, Volume XXXVI, ISSN No , Hannover, Germany, May 17 to 20, 2005., 071-pernar.pdf, 4 pages. [11] Schaepman, M., et al., Experimental determination of the adjacency effects over an eutrophic lake using a helicopter mounted spectroradiometer for the correction of imaging spectrometer data, Third International Airborne Remote Sensing Conference and Exhibition, 7-10 July, 1997, Copenhagen, Denmark. [12] Schlaepfer, D., PARametric Geocoding, Orthorectification for Airborne Scanner Data, User Manual Version 2.3, ReSe Applications Schlaepfer and Remote Sensing Laboratories (RSL) of the University of Zurich. [13] Šemanjski, S., Gajski, D., Gold, H., GPS aided INS Integration and Application in the Croatian Sky, 1st GNSS Vulnerabilities and Solutions Conference, Baška, Krk Island, Croatia, 7 9 September, 2008.