Date of Defense
16-4-2025 5:00 PM
Location
F1-2010
Document Type
Thesis Defense
Degree Name
Master of Science in Electrical Engineering (MSEE)
College
COE
Department
Electrical and Communication Engineering
First Advisor
Dr. AbdulHalim Jallad
Keywords
Lossless compression, near-lossless compression, parallel implementation, FPGA, hyperspectral images, near-lossless compression, remote sensing.
Abstract
Hyperspectral images are used in remote sensing scientific research and more, but they can reach hundreds of megabytes in size. This large number of datasets creates a problem with the limited storage capacity and bandwidth available on the satellite which in turn necessitates the use of hyperspectral image compression algorithms. The near lossless CCSDS123 hyperspectral algorithm provides a high compression capability, but it has some data dependencies in its predictor module that slows down the compression process. This Master of Science (MSc) thesis focuses on implementing a parallel, real-time FPGA based architecture of the near lossless CCSDS123 compression standard. The objective of this thesis is to compare the lossless and near lossless versions of the standards, study the parallel FPGA implementation of the CCSDS123.0-B-2 in terms of image compression performance and resource utilization trade-offs, and benchmark it to both lossless and near lossless CCSDS123 state of the art implementations. The implementation was tested on Zybo Z7 board containing a Zynq-7020 FPG. It achieved a throughput of 40.008 Msamples/s due to the critical path in the quantizer mapper and a moderately low power of 1.586 W. The implementation is realtime, far exceeding the AVIRIS’s throughput of 20.4 Mbits/s by achieving a throughput of 640.128 Mbits/s. This thesis highlights the tradeoffs between image compression performance and resources utilization for near lossless CCSDS123, helping in the development of efficient hardware designs in future.
Included in
A PARALLEL, REAL-TIME FPGA IMPLEMENTATION OF THE CCSDS 123.0-B-2 STANDARD
F1-2010
Hyperspectral images are used in remote sensing scientific research and more, but they can reach hundreds of megabytes in size. This large number of datasets creates a problem with the limited storage capacity and bandwidth available on the satellite which in turn necessitates the use of hyperspectral image compression algorithms. The near lossless CCSDS123 hyperspectral algorithm provides a high compression capability, but it has some data dependencies in its predictor module that slows down the compression process. This Master of Science (MSc) thesis focuses on implementing a parallel, real-time FPGA based architecture of the near lossless CCSDS123 compression standard. The objective of this thesis is to compare the lossless and near lossless versions of the standards, study the parallel FPGA implementation of the CCSDS123.0-B-2 in terms of image compression performance and resource utilization trade-offs, and benchmark it to both lossless and near lossless CCSDS123 state of the art implementations. The implementation was tested on Zybo Z7 board containing a Zynq-7020 FPG. It achieved a throughput of 40.008 Msamples/s due to the critical path in the quantizer mapper and a moderately low power of 1.586 W. The implementation is realtime, far exceeding the AVIRIS’s throughput of 20.4 Mbits/s by achieving a throughput of 640.128 Mbits/s. This thesis highlights the tradeoffs between image compression performance and resources utilization for near lossless CCSDS123, helping in the development of efficient hardware designs in future.