Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform
Twilley, R. 2025. Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/ad517ab4-97ad-4455-b497-045e676332c4
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Stellenbosch : Stellenbosch University
2025
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867614004973666304 |
|---|---|
| access_status_str | Open Access |
| author | Twilley, Rowan |
| author2 | Barnard, Arno |
| author_browse | Barnard, Arno Twilley, Rowan |
| author_facet | Barnard, Arno Twilley, Rowan |
| author_sort | Twilley, Rowan |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Twilley, R. 2025. Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/ad517ab4-97ad-4455-b497-045e676332c4 |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/132387 |
| institution | Stellenbosch University (South Africa) |
| language | English |
| last_indexed | 2026-06-10T12:45:08.467Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/132387 Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform Twilley, Rowan Barnard, Arno Stellenbosch University. Faculty of Engineering. Dept. of Electrical & Electronic Engineering. Remote-sensing images Real-time data processing Remote sensing -- Data processing Computer hardware description languages Metal oxide semiconductors, Complementary Programmable logic devices UCTD Twilley, R. 2025. Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/ad517ab4-97ad-4455-b497-045e676332c4 Thesis (MEng)--Stellenbosch University, 2025. ENGLISH ABSTRACT: This thesis addresses onboard image correction needs for satellite imagers by implementing synthesizable HDL algorithms for inline data processing. The goal is to apply pre-calculated correction coefficients to compensate for changes in camera optics and front end electronics. By performing image correction on raster pixel data during image acquisition, images can be corrected using fewer memory resources and less power. The solution takes the form of modelling, from first principles, the noise that is present in Complementary Metal Oxide Semiconductor (CMOS) image sensors that capture images using the line scan imaging technique. Furthermore, geometric lens distortion was modelled with consideration for the physical relationship between the lens and image sensor. The models produce simulated raw satellite images that characteristically mimic real raw satellite images. Pipelined synthesizable Hardware Description Language (HDL) algorithms were designed and implemented for application on a Field Programmable Gate Array (FPGA). Two correction algorithms were developed as solutions: gain-offset correction, using two-point normalization, and geometric lens distortion correction. This utilizes pixel movement vectors based on characterized distortion profiles. Both algorithms operate inline, between signal digitization and storage of pixel data, on continuous pixel data stream. The HDL algorithms were end-to-end tested and benchmarked on the PolarFire™ Discovery Kit. The challenges involved in algorithm testing on a real satellite led to experimental tests in which raw image data from the models were uploaded and streamed to the FPGA, thereby simulating the line scan image capture. The accuracy of the corrected image is evaluated using Peak Signal-to-Noise Ratio (PSNR), Mean Absolute Error (MAE), and Structural Similarity Index (SSIM). The algorithm’s performance is assessed based on FPGA resource utilization and processing speed. The developed algorithms achieved a uniformity score exceeding 97%, with PSNR improvements of 30 to 40 dB for intensity correction while processing at 18.87 Megapixels per second. Geometric distortion correction reduced the mean absolute error (MAE) by over 50% and improved the structural similarity index (SSIM) by 39.5% in cases of high distortion, achieving SSIM values between 0.8 and 0.9, while operating at 125 Megapixels per second. These results highlight the effectiveness of the proposed FPGA-based correction approach in improving image quality in satellite imaging systems. AFRIKAANSE OPSOMMING: Hierdie tesis fokus op die behoefte aan aanboord beeldkorreksie vir satellietbeelde deur sintetiseerbare HDL-algoritmes te implementeer vir inlyn-dataverwerking. Die doel is om voorafberekende korreksiekoeffisiente toe te pas om vir veranderinge in kamera-optika en voorkant-elektronika te kompenseer. Deur beeldkorreksie tydens beeldverkryging op rasterpikseldata toe te pas, kan beelde reggestel word met minder geheuegebruik en laer kragverbruik. Die oplossing behels die modellering van geraas in Complementary Metal Oxide Semiconductor (CMOS) beeldsensors, gebaseer op eerste beginsels. Hierdie sensors neem beelde op deur ’n lyn skandering beeldtegniek te gebruik. Verder is geometriese lensvervorming gemodelleer deur die fisiese verhouding tussen die lens en beeldsensor in ag te neem. Die modelle genereer gesimuleerde rou satellietbeelde wat ooreenstem met werklike rou satellietbeelde. Gesintetiseerde gepyplynde Hardware Description Language (HDL) algoritmes is ontwerp en geimplementeer vir gebruik op ’n Field Programmable Gate Array (FPGA). Twee korreksie-algoritmes is ontwikkel: gain-offset-korreksie deur tweepunt-normalisering, en geometriese lensvervormingskorreksie wat pikselbewegingvektore gebruik, gebaseer op gekarakteriseerde vervormingsprofiele. Beide algoritmes werk inlyn, tussen digitalisering en die berging van pikseldata op ’n deurlopende pikselstroom. Die HDL-algoritmes is volledig getoets en vergelyk op die PolarFire™Discovery Kit. Die uitdagings van toetsing op ’n werklike satelliet het gelei tot eksperimentele toetse waar rou beelddata vanaf die modelle opgelaai en na die FPGA gestroom is, wat die lyn skandering beeldvasleggingsproses nageboots het. Die akkuraatheid van die gekorrigeerde beeld is ge¨evalueer met behulp van Peak Signal-to-Noise Ratio (PSNR), Mean Absolute Error (MAE) en die Structural Similarity Index (SSIM). Die werkverrigting van die algoritmes is beoordeel op grond van FPGA-hulpbronbenutting en verwerkingspoed. Die ontwikkelde algoritmes het ’n eenheidskoers van meer as 97% behaal, met PSNRverbeterings van 30 tot 40 dB vir intensiteitskorreksie teen 18.87 Megapixels per sekonde. Geometriese vervormingskorreksie het die gemiddelde absolute fout (MAE) met meer as 50% verminder en die SSIM met 39.5% verbeter in gevalle van ho¨e vervorming, met SSIM-waardes tussen 0.8 en 0.9, teen ’n 125 Megapixels per sekonde verwerkingspoed. Hierdie resultate beklemtoon die doeltreffendheid van die voorgestelde FPGA-gebaseerde korreksiemetodes om beeldkwaliteit in satellietbeeldverwerking te verbeter. Masters 2025-06-05T13:42:09Z 2025-06-05T13:42:09Z 2025-03 Thesis https://scholar.sun.ac.za/handle/10019.1/132387 en Stellenbosch University xviii, 147 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Remote-sensing images Real-time data processing Remote sensing -- Data processing Computer hardware description languages Metal oxide semiconductors, Complementary Programmable logic devices UCTD Twilley, Rowan Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform |
| title | Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform |
| title_full | Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform |
| title_fullStr | Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform |
| title_full_unstemmed | Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform |
| title_short | Design and implementation of real-time on-board satellite image correction algorithms on a programmable logic platform |
| title_sort | design and implementation of real time on board satellite image correction algorithms on a programmable logic platform |
| topic | Remote-sensing images Real-time data processing Remote sensing -- Data processing Computer hardware description languages Metal oxide semiconductors, Complementary Programmable logic devices UCTD |
| url | https://scholar.sun.ac.za/handle/10019.1/132387 |
| work_keys_str_mv | AT twilleyrowan designandimplementationofrealtimeonboardsatelliteimagecorrectionalgorithmsonaprogrammablelogicplatform |