Real-Time Vehicle License Plate Detection and Tracking Using Multilayer Back-Propagation Neural Networks with and without Hough Transform

Abstract

There are many applications of vehicle tag detection systems using image properties, which include image density, corner detection and blob analysis but using neural networks in real time for both detection and tracking is always challenging in terms of the computational load and the quality of the input images for training purposes. In this case, Hough transform is found to be useful to extract the square shaped vehicle tag from the scenario in order to improve the output of the neural network regardless of the different background scenarios.

In this study, first real time detection and tracking of a specific tag is examined using Multilayer Back-Propagation Neural Networks, its high computational load and background noise dependency, the effect of the quality of the input samples on the accuracy of detection and tracking are proved. Then in chapter two, a new approach, using Hough Transform in parallel with Neural Networks is simulated both in MATLAB and RoboRealm.