Remote Sensing (RS) data obtained from satellites and aerial flights are valuable source of information providing up-to-date information about natural resources like land, water, forests, urban areas and infrastructure facilities like transportation network, canal network etc. Geographical Information Systems (GIS) are widely being used for management, monitoring and planning of natural resources. In order to make this information available to the GIS database, the image containing RS data has to be integrated with the GIS, it is possible to make use of these data by a GIS in terms of spatial data layers updation, analysis of spatial data with attributes in conjunction with RS data and generation of maps containing combination of RS image and other types of cartographic and thematic information.
This paper discusses and describes requirements and procedure for digital image integration into GIS databases. The issues related to image integration are discussed and a procedure of geo-referencing image within the GIS database framework, GIS database layer updation and composition and plotting of map containing image data as well as selected database layers has been described in detail with special reference to Natural Resources Information System (NRIS). Ultimately, this approach will lead to development of a user friendly software Shell for carrying out the tasks of image registration with spatial databases (e.g. NRIS), generation of image-maps by overlaying selected database elements at desired scales and paper sizes and updation of existing spatial database layers like landuse, transportation network, drainage, canal network etc. Towards this, a prototype is under development, which conforms to NRIS database standards.
Image Integration with Spatial Databases
Issues Related To Image Integration
A satellite image or aerial image can not be integrated as it is with a spatial database because of two reasons. The first reason is that a satellite image suffers from geometric distortions because of platform instabilities and attitude effects and second reason is that the coordinate system of the image and the node database are different. Because of the different co-ordinate systems, an image can not be displayed along with a spatial database layer because their extents do not match. Due to geometrical distortions, features of the image do not match with the spatial database features because of differences in scale, translation and rotation effects. Unless these issues are resolved, image integration can not be performed.
Geo-referencing
An image can be integrated with the spatial databases using geo-referencing technique. Using this technique geometrical distortions are removed from the image and the co-ordinate system of the image is changed to the co-ordinate system of the database. Geo-referencing is done by establishing links between image features and corresponding features in the spatial database. Features used for establishing links are called Ground Control Points (GCPs) which should be clearly identifiable on the image as well as on the spatial database. From the GCP positions on the image and spatial database, the image is geo-referenced by making use of a mapping polynomial. This mapping procedure removes geometric distortions of the image and changes the co-ordinate system of the image to spatial database co-ordinate system. Normally, the geo-referencing is done in interactive mode. The user identifies GCPs interactively on the image and database and establishes links. Based on these links geo-referencing process calculates mapping polynomial coefficients and image is mapped with the spatial database. However, this process could be made automatic if image corner points are known in terms of latitude and longitude values. Since no user interaction is required, this geo-referencing technique could be called automatic geo-referencing.
The process of image integration consists of three steps. These steps are image extraction, automatic geo-referencing and interactive geo-referencing. The steps required for image integration are described below.
Image Extraction
The process of image extraction involves extraction of the image data from the DAT/CD media and storing in the disk file, preparation of the image header file containing information about image size, number of bands etc., preparation of image statistics file giving information about minimum and maximum gray levels, mean and standard deviation for each band and extraction of latitude-longitude information for all four image corners into a file from the ancillary information provided on the media. The image extraction program can generate these four files which are used in subsequent steps.
Automatic Geo-referencing
After image extraction, the next step is to perform automatic geo-referencing of the extracted image using satellite ephemeris information available in form of corner points latitude-longitude information and line number-pixel number information. During automatic geo-referencing the following steps should be performed:
- Conversion of the corner points latitude-longitude information to the node database map co-ordinate system.
- Generation of image corner points data file containing pixel number, line number and their co-ordinates in the map projection system for each corner point.
- Generation of coefficients file for mapping polynomial by making use of the image corner points.
- Automatic geo-referencing of the extracted image using mapping polynomial.
The process of automatic geo-referencing changes the co-ordinate system of the image to that of spatial database and removes the geometric distortions up to the accuracy of the ephemeris information, which could be of the order of several meters or a few kilometers. Since the automatic geo-referencing process uses only first degree mapping polynomial, higher order distortions other than translation and scaling are still not removed. In order to remove the higher order distortions left in the image and register the image as accurately as possible with the spatial database layers, the automatically geo-referenced image is still required to be further geo-referenced using interactive geo-referencing process.
Interactive Geo-referencing
The objective of the interactive geo-referencing is to remove geometric distortions present in the auto geo-referenced image due to inaccuracy of the ephemeris information. The major steps for interactive geo-referencing are as follows.
- Identification of links between image and spatial database features.
- Evaluation of links identified in terms of Root Mean Squared (RMS) error between positions of links in the spatial database and transformed positions of links in the image.
- Geo-referencing the image using the evaluated links and a higher order mapping polynomial to register the image more accurately with the database layers and storing the geo-referenced image into an image catalogue.
Image Catalogue Creation An image catalogue is a logical collection of images normally belonging to an area of interest. The image catalogue contains information about the images stored in the catalogue as well as other information about the image such as date of image data acquisition, sensor name, geometric resolution of the image, geographical extents of the image etc. After image integration is done, the images could be stored in the image catalogue for future use.