Model based implementation of a real-time image processing system

Image processing systems are frequently used in many fields today. A significant portion of these systems have to work in real time due to the nature of system. Because of the need for real-time operation on such systems requires high processing power, it is possible to use application-specific embedded hardware and software designs. For these reasons, the Field Programmable Gate Arrays (FPGA) are often used in embedded system designs that will work in real-time. In recent years, with the development of semiconductor technology, the use of FPGA and microprocessor systems in the same integrated circuit has been presented. Systems that are fully programmable and have much lower power and space consumption than classical systems can be developed in this way. However, a major disadvantage of such systems is the need for both hardware and software design, and advanced design knowledge and experience in hardware design. High-level synthesis and design tools are developed to overcome this situation which can be used to design hardware without using hardware description languages. In this thesis study, a real time image processing system is designed by using model based design methods in Matlab / Simulink environment which is a new generation high level programming tool with Xilinx Vivado Design Suite. Zedboard development board with Xilinx Zynq-7000 architecture is used in the study. The processed image is transferred to the monitor by applying edge detection, noise reduction, grayscale conversion and sharpening filters on the image taken from the camera instantaneously. The Zynq platform used in this study is an integrated system architecture which consists both microprocessor and FPGA what makes it different from the classical FPGA architecture. Because of that, the system architecture and the model based design methods are given in detail first. The following sections provide information on image processing methods and how they are applied to real-time video. Then, these image processing and filter algorithms were designed and simulated with model-based tools in Matlab / Simulink environment, then hardware integration and software design of the image processing system designed in Xilinx development environment were implemented and system usage was analyzed on Zedboard. As a result of the studies done, it has been observed that the system worked successfully in real time.