Detection of the content resembling character sequences on images with gabor filters

Because interaction relationship between human and machine is increasing, new algorithms are developed. Via these algorithms, machines are won artificial intelligence and human interaction increases. Image processing techniques have an efficient role in human machine intereaction. Many problem can be solved with image processing techniques and these problems includes character series on images. If it is illustrated, plate location recognition problems can be one of the best example. In the literature, a lot of publications about plate location recognition. Different methods are used such as edge detection. Even if recognition rate is high of these methods, the algorithms have difficulties by implementation on numerical systems such as FPGA. If these problems are considered, a fast recognition algorithm can be designed by obtaining oriented components of plates using Gabor-type filters which are designed with Cellular Neural Networks (CNN). CNN Gabor-type filters are implemented effectively in hardware of FPGA in many literature works. Thus, algorithms, which have high recognition rate can be developed benefiting from features of selection of direction and frequency of Gabor-type filters. In this study, an algorithm, which is convenient for digital design and it has 81 % accuracy rate is developed on plate location recognition. In this method, CNN Gabor-type filters are used and morphological operations such as Gabor-type filters direction selector structure is used.