Abstract:
Field-programmable gate arrays (FPGAs) have become a popular choice for high-
speed data transmission systems due to their high performance and flexibility. FPGA-
based systems are increasingly being used in various applications, such as radar, medical
imaging, autonomous driving, and quantum computing. Within this paper, a detailed
examination is presented, offering insights into FPGA-centric image and video
processing, accompanied by a thorough exploration of the associated design
considerations. The FPGA used in this thesis is the ZCU104. The advantages of FPGA-
based systems, including high-speed data processing, low latency, and reconfigurability
are discussed. The paper also emphasizes certain challenges and limitations associated
with the design of FPGA-based data transmission systems. To illustrate the practicality
of FPGA data processing, PYNQ has been used as it contains Open Computer Vision
Library (CV). The main focus has been implementation and comparison of different
methods used for image and video processing. In order to measure the performance of
these algorithms, several aspects have been taken into consideration such as accuracy or
frame rate. In addition, the limitations of this hardware have been discussed. The aim of
this thesis is analyzing the performance of these processing algorithms and what can be
done for future improvements. Overall FPGA are an excellent choice compared to the
traditional approaches.
