Projects / Programmes
Automatic generation of true orthoimages from very high-resolution data
| Code |
Science |
Field |
Subfield |
| 2.17.00 |
Engineering sciences and technologies |
Geodesy |
|
| Code |
Science |
Field |
| T181 |
Technological sciences |
Remote sensing |
| Code |
Science |
Field |
| 2.07 |
Engineering and Technology |
Environmental engineering
|
automatic processing, satellite images, orthorectification, true orthoimage, high-resolution surface model, geometric modelling
Researchers (1)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
28658 |
PhD Aleš Marsetič |
Geodesy |
Head |
2017 - 2019 |
107 |
Organisations (1)
Abstract
Remote sensing data is becoming an increasingly important source for supplementing existing geospatial data, or even upgrading it. Due to the high quantity of new and existing data the automation of its processing has become very important, as only (semi)automatic processes can provide large quantities of end products in near real time.
In this respect, a special attention should be given to the very high-resolution (VHR) optical satellite data. Its best resolution (currently 0.3 m) is already comparable to the resolution of aerial images, usage application areas are almost identical. However, since no automated procedures for their processing that would provide sufficiently accurate and high-quality products are available, their potential is still not completely exploited.
Before optical satellite imagery can be used for spatial analysis, combined with a variety of other geo-referenced data, they have to be orthorectified into so-called orthoimages. Orthoimages have a high usability, since they contain both image characteristics of a photograph and geometric properties of a topographic map. Nowadays orthoimages are indispensable in particular as a raster basis for a variety of geospatial applications. They can replace or complement vector data or generate geodetic data for the purpose of planning.
True orthoimages are an improved version of traditional orthoimages as they are adjusted also for high objects (e.g. high buildings). Generation of true orthoimages is more complex than the generation of usual orthoimages, since it requires more images of the same area, acquired from different angles (the so-called image block).
The aim of the proposed research project is to develop an automatic procedure for orthorectification of VHR satellite images, and generation of traditional orthoimages or true orthoimages in a very short time. The results can be used directly by end-users for the production of spatial data or in GIS analyses.
The proposed project is organized in five work packages (WPs):
WP1: Automatic extraction of ground control points (GCPs) from VHR images,
WP2: Geometric model for satellite images processing,
WP3: Direct orthorectification for the generation of true orthoimages,
WP4: Generation of true orthoimages, and
WP5: Automation, optimisation and validation of developed procedures.
In WP1 we will develop an automated two-step procedure for the extraction of GCPs where digital roads and aerial orthophotos will be used as reference data. WP2 will be devoted to the geometric model, which will calculate the orthorectification parameters for single image or image block. In WP3 we will analyze the methods for orthoimage generation. Special attention will be devoted to the generation of a high-resolution surface model and the methods of projecting the image coordinates to the orthoimage. WP4 will be devoted to the generation of a true orthoimage. The procedures for the occlusion filling, radiometric balancing and final mosaicing will be developed. In WP5 we will integrate the procedures developed in previous WPs into a single final workflow for the automatic generation of ortoimages from VHR data.
The proposed project will deal with a research topic that will bring many innovations in the field of remote sensing and will also have a big potential for application in other scientific fields. The extensive analyses that are going to be performed will enable the transfer of knowledge to actual systems or databases that will be used by potential public or private end-users. The project is harmonized with the needs and demands of current project calls, published by European Space Agency and European Union.
By achieving the project goals, we expect significant advances when compared to the existing approaches in the field of automatic true orthoimage generation from satellite data. With the specific knowledge gained, the Slovenian expertise will contribute to the development of the newest research in this field.
Significance for science
The review of the literature indicates that the field of automatic generation of (true) orthoimages is a subject of intensive research, but is still far from established solutions. This is especially true when the orthoimages are generated from very high-resolution (VHR) satellite images, acquired from different angles.
The presented method for automatic generation of true orthoimages open up a wide range of tools for satellite image processing. In addition to the traditional and true orthoimages, also other results, such as terrain and surface models or 3D reconstruction of objects, can be generated from a block of images.
The proposed project plausibly continues the work and enhances the results completed during the doctoral dissertation. The automatic procedure, which was fully implemented in the programming language IDL and included into the STORM processing chain, will be expanded with new procedures for VHR satellite images. The project will result in the following new methods:
an advanced method of image matching that will provide a high absolute accuracy of the orthoimages,
enhanced geometric model that will compute (without operator’s intervention) a set of optimal parameters for the orthorectification that will be based on the number and accuracy of the ground control points, and
a method for automatic generation of orthoimages or true orthoimages using an upgraded lidar surface model and utilizing a direct orthorectification method.
With the developed procedures the automatic generation of a more accurate VHR orthoimage or true orthoimage will be possible. Both types of orthoimages are useful in many applications that deal with spatial data. Furthermore, the automatization of processes is necessary in situations where there is a strong need for rapid response and/or if the amount of data is too large for manual processing. For this reason in an era of abundance of digital data the design and implementation of the automatic procedure will be useful also in other areas.
The development of the proposed procedure is fully consistent with the expansion plans of the STORM processing chain that is being developed within the research group for many years. At the end of the development the proposed procedure will be integrated into the STORM processing chain as one of the processing modules. With the proposed procedure we will be able to follow the global trends in VHR image processing and contribute to the progress of this segment of remote sensing.
Due to the quantity and nature of available data and considering experiences and results the project applicant has in remote sensing, we expect the research to result in important new findings within the specific research area. New methodology will offer wider implications for the development of specific methods and will satisfy the needs, expectations and requirements of the end-users, which will be the target group when developing this system.
Significance for the country
The research performed in this project will significantly improve the planar accuracy of orthoimages and with the automation of all processes it will extremely accelerate the process of their production, especially in cases when repeated observations are needed (e.g. regular monitoring of a specific event). Even the current version of the geometric corrections in the STORM chain has been used various times as a service by the Ministry of Agriculture, Forestry and Food for the need of data management in agriculture. With the improvement of the proposed method and with a better quality of its products the usefulness of the service will be even greater - it will attract also end-users from the private sector, as well as from abroad.
Satellite orthoimages can also complement existing aerial orthophotos (i.e. national orthophoto, DOF) used by both public institutions and private companies. In Slovenia, aerial DOF is acquired in the frame of cyclic aerial survey. The whole country is covered by DOF with a spatial resolution of 0.25 m every three years. In the coming years the satellite orthoimages will not replace the aerial orthophotos (due to poorer spatial resolution and price), but they could be used as a useful complement, as satellites can achieve significantly higher temporal resolution (revisit time can even be a few days) and may quickly cover the desired area (for the entire Slovenia a few dozen images are enough). This enables accurate monitoring of various phenomena on the Earth's surface. Any such monitoring involves a large amount of satellite data, which can be processed, within a reasonable time and financial capability, only with the proposed automatic procedure.
In the coming years true orthoimages will replace the traditional orthoimages and become a standard reference data of most high developed countries. The procedure that is defined in the frame of the project will considerably shorten the time of processing compared to existing algorithms for generation of true orthoimages from VHR satellite images. With some adaptations the procedure could be used also for aerial imagery.
The results of this study would be useful for preparing the standards for national or local level mapping using true orthoimages and for the establishment of very high-resolution satellite images (and derived true orthoimages) as a new source for the acquisition of national topographic data and updating topographic databases.
The results of the procedure will enable faster interpretation of large quantities of VHR images to obtain spatial information for the needs of various disciplines involved in the study of changes in the Earth's surface, environmental and cultural heritage management and mitigation of natural hazards, as well as historical analysis of landscapes and its anthropological modifications. They will also be useful in most present and future projects within the research group and in the current ARRS research programme titled Anthropological and spatial studies.
Most important scientific results
Interim report,
final report
Most important socioeconomically and culturally relevant results
Interim report,
final report
