الفهرس 
fussy image processingOnly 14 pages are availabe for public view
 |  | from | 208 |  |  |
| | | from | 208 | | |
Abstract
The tremendous advancement in modern techniques of digital image processing has led to a massive change in most of scientific and practical fields. Whereas the applications are extended depend on digital image processing.
Many difficulties in image processing arise due to the data, tasks, and results are uncertain. This uncertainty, however, is not always due to the randomness but to the ambiguity and vagueness. These problems are fuzzy in nature making that the fuzzy approaches are the suitable way to manage these imperfections.
Reconfigurable hardware in the form of Field Programmable Gate Arrays (FPGAs) has been proposed as a way of obtaining high performance for Image Processing, even under real time requirements.
Implementing image processing algorithms on reconfigurable hardware minimizes the time-to-market cost, enables rapid prototyping of complex algorithms and simplifies debugging and verification. Therefore, FPGAs are an ideal choice for implementation of real time image processing algorithms.
In this thesis, two approaches of General Purpose Fuzzy Processor (GPFP) are designed and implemented on FPGA to be fuzzy engines for different applications especially for real-time fuzzy image processing applications.
The first approach, is an intuitive method and is based on the calculation method in every stage of fuzzy processor and is designed to be faster than the processor (Pentium 4, 3 GHz) by 30 times. Although this speed, it is not suitable for the real-time fuzzy image processing.
2/2
The second approach is a reasonable approach and is based on Memory Look-Up Table (MLUT) concept and designed to accomplish the required speed for image processing applications.
VHDL language is used to describe the performance of the proposed fuzzy processor and Xilinx ISE is used for synthesis and implementation of the circuit on Spartan-3E starter kit.
XtremeDSP Spartan-3A Video Starter kit is used to verify the implementation of fuzzy-based edge detecton and skin detection using the proposed GPFP.
Both approaches of the proposed fuzzy processor are designed to be suitable for up to four inputs and one output, since most of the applications which uses the fuzzy systems don’t need more than these input and output specifications.
The required parameters for both of the proposed GPFP approaches are prepared and tested by C# program is designed for this purpose.
The proposed GPFP approaches are tested through four different real-time applications with different parameters to all designed approaches to verify the speed and the functionality of each approach and check the compatibility for image processing applications.
Experimental results of the proposed GPFP approaches show that the first approach failed to verify the requirements for fuzzy image processing applications but it gave excellent performance when is used as fuzzy controller applications, and the second approach succeeded in the usage of real-time fuzzy image applications, and both approaches are considered to be ready for the usage of any real-time fuzzy applications compatible with their specifications.