Recognition of Moving Vehicle Number Plates using Convolutional Neural Network and Support Vector Machine Techniques

Abstract

Nowadays video cameras have become gradually deployed, hence the hassle of video enhancement has also been increased. Video enhancement is a process of illuminating the occurrence using gentle techniques to maintain the integrity of pixel quality. The standard of the original video recording gives the success for the enhancement. The purpose of video enhancement is to refine the visual look of the video or to give an extra changed illustration for future video processing which consists of analysis, detection, segmentation, recognition, and used for surveillance and the criminal justice system. In the proposed work vehicle number plate is enhanced and recognition of a number plate is performed using Convolutional Neural Network and Support Vector Machine. There are a lot of challenges in recognizing the number plate due to the presence of blur, low-intensity, snow, rain, hit and run cases. In such a case, recognizing the vehicle number plate is challenging. So to overcome all these problems video enhancement has to be performed. The proposed work involves converting the video into image frames, pre-processing the frames and then performing enhancement, and finally recognizing the vehicle number plate using CNN and Support Vector Machine. The result analysis proves that CNN gives better classification accuracy over the Support Vector Machine model.