Georeferencing on Synthetic Aperture RADAR imagery |
Paper ID : 1029-SMPR-FULL |
Authors: |
مجید اسماعیل زاده * خیابان کارگر شمالی - بعد از بزرگراه جلال آل احمد - کوی دانشگاه تهران |
Abstract: |
Geocoding accuracy of Synthetic Aperture RADAR images plays a key role in many aspect such as large-scale high resolution mapping, multi-source data fusion and other quantitative applications. SAR images are generated through signal processing using the data collected by the SAR system from the target scenes on earth surface. These SAR images include errors or distortion which cause images to be inaccurate. These errors in SAR images called geometric distortions. Geometric distortion are inherent in the uncorrected image data they can generally be categorized as resulting from sensor instability, platform instability. Signal propagation effect, terrain height and processor induced errors. The propose of this research is to reduce the error of SAR images through implementation of geometric correction based on Range-Doppler model. Geometric correction or equivalently geometric rectification refers to the process step where the image is resampled to new projection. Geocoding usually refers to a spatial case if the geometric correction procedure where the image is resampled to some spatial representation with known geometric properties (e.g., a standard map projection such as Universal Transverse Mercator, UTM). The three main process of geometric correction are mathematical model, coordinate registration and resampling. In previously research, to obtain high accuracy traditional method is using Ground Control Points (GCPs) or digital terrain data to make geometric correction these methods are not suitable because does not measure and compensate for each error source but eliminate the whole errors by mathematic approach.so, its precision heavily depends on the number and distribution of GCPs or the accuracy of digital terrain data. Lately, Rational Function Modeling (RFM) is proposed and gives a good result and has fine accuracy but has some limitation. This can be accomplished by employing the terrain-independent approach or terrain-dependent approach which depends upon the availability of physical sensor models. In terrain-independent approach you need physical sensor model attitude and in terrain-dependent approach its need sufficient GCPs evenly distributed on the observed terrain surface and these are limitation of this method. In this paper, we introduce an approach that does not need any GCPs or physical sensor model. This method uses satellite precise ephemeris data for georeferencing and employs DEM to compensate errors induced from topography height such as foreshortening. This method willing a DEM that SAR image scene exists on it. Now it’s time to create simulated image and transfer DEM to SAR geometry in range and azimuth, it means that every point in DEM has coordinate in range and azimuth system. For this procedure, DEM points must transferred from geodetic coordinate system to geocentric coordinate system because ephemeris data that presents satellite position in space are on geocentric system. To find value of range and azimuth, a mathematical model needs and using Range-Doppler model is suitable. In the RD model, the geometric characteristics of SAR imaging procedure are described by two equation: the slant range equation (1) and Doppler equation in azimuth dimension (2) R=| R_s-R_t | (1) f_D=2/λR(V_s-V_t) .(R_s-R_t) (2) Which S means satellite and T means target. Using these equation and finding the minimum distance between each point in DEM and satellite position, the place of DEM points in range-azimuth system can be calculated. Now, there is a DEM that every point on this has coordinate in geodetic and range-azimuth system. Before this process, computations applied for correct topography due to the difference between the local radius and center radius to compensate foreshortening error. After that, the position of SAR image into the simulated DEM detects and the original SAR image can be register to simulated DEM. The final product of this method is georeferenced TIFF image in geodetic coordinate system. The planimetric accuracy is less than 2ʺ and elevation has accuracy about 10 meters. |
Keywords: |
Georeferencing-Digital Elevation Model- Precise Ephemeris |
Status : Paper Accepted (Poster Presentation) |