Handwritten character recognation system desing for digital hardware implementation

Interaction relationship between human and machine is increasing in modern-day gradually. The machines are acquired artificial intelligence by improved new algorithms and current autonomous systems facilitate various services in the human daily life. Image Processing technique are intensively used for developing this systems. A various problems are solved by using designed autonomous structures. Handwriting recognition, which is the subfield of optical characters recognition, is one of the major of the problems. There are successful algorithms on handwriting recognition according to recognition rate in technical literature. Improved algorithm has also high recognition rate however the algorithm has difficulties by implementation onFPGA or other hardware because of contenting square root, division operations which get more load of mathematical processing. According to this specific problem, a fast recognation algorithm can be designed by obtaining oriented components of letters using Gabor type filters in cellular neural networks (CNN). CNN gabor type filters are implemented in many literature works efficiently. By this means, fast recognation rate algorithms are designed with obtaining oriented components feature of Gabor filters. On the other hand, oriented components could be obtained from the letters by morphological operations such as Gabor type filters direction selector structure considering the images which is processing are binary. In this study, two different algorithms which is convenient for digital design and has 80% to 91% accuracy rate are improved on handwriting recognation. In first method CNN gabor type filters are used, in other methods morphological operations such as Gabor type filters direction selector structure is used. Keywords: handwriting character recognition, cellular neural Networks, Gabor filter, Morphological filter