Cellular neural networks and orientation selective Gabor filters

Nowadays, the Artificial Neural Network discovered being inspired by the nature has been reduced to practice immensely in many different disciplines. In 1988, the Cellular Neural Network (CNN) architecture was introduced by L. O. Chua and L. Yang whose aim was to model the processing of mankind nervous systems. Because of the analog cellular structure, this new architecture has brought in new point of view for applications as image processing that experience high density of computing and long response time bottleneck when using digital systems. The CNN that yields results as short as an analog system?s response time can be used widely in image processing application. The analysis of the specific information on the signals can be achieved successfully with the Gabor filters inspired by Quantum Theory in 1946. The 2-D Gabor filters are preferred in image processing field by reason of the orientation and frequency selectivity characteristics. The density of computing which depends on the input and filter?s size at the signal processing application shifts scientist?s attention to find alternative methods that respond more quickly. The Recursive Gabor filters and the CNN Gabor-Type filters are two of the alternative methods that discovered for Gabor Filters. In this study; after giving basic concept and definition for the Cellular Neural Network, the availability of the CNN for image processing filed has been showed with a practice example. The Recursive Gabor filter method has been examined with comparative application after particularizing the Gabor filters and its importance in the image processing field. The CNN Gabor-Type filter has been given with its analog circuit?s implementation and the Elliptical Gabor Filters has been defined as an original structure. The MATLAB 7.4 programming language has been used for all application in this thesis. Keywords: Cellular Neural Network (CNN), Gabor, Filter, Image Processing, Recursive, Elliptical.