A Hybrid Technique for Real Time License Plate Localization And Recognition With The Aid of Ffbnn Apso

Abstract

ABSTRACT newline newline newlineIn this research work, a general approach for the problem of Automatic Vehicle License Plate Recognition (AVLPR) system is proposed and is deliberated. At this juncture, an efficient AVLPR system for identifying the registration number of a License Plate (LP) with Indian text style is designed and is premeditated. Accordingly, numerous algorithms have been projected in recent period for a proficient exploitation of such meticulous applications. Now, this proposed AVLPR system offers an uncomplicated innovative conception and a perceptible blend of certain schemes and a couple of innovations for the localization and characters identification, which turns to be a hybrid system for Indian LPs. The positioning and extraction of LPs from the automobile pictures and the characters distinguishing and identification within each LP, are the basic modules of a general AVLPR system. newlineThe proposed algorithm establishes soft computing techniques entrenched in fuzzy rule-based classifier, Adaptive Neuro Fuzzy Inference System (ANFIS) for License Plate Detection (LPD) phase and Neural Network (NN) discipline, Feed Forward Back-Propagation Neural Network (FFBNN) for License Plate Character Recognition (LPCR) phase, subsequent to the features extraction for both the LPs and the LP characters; in order to compensate the uncertainties, noisy like problems, illumination difficulties, measurement faults and insufficient progresses within the existing systems. The Adaptive Particle Swarm Optimization (APSO) Genetic Algorithm (GA) is implemented both with ANFIS and FFBNN, beneficial to optimize the parameters throughout the training practice, so as to enhance the processing period and accurateness. newlineThe projected AVLPR system employs mostly nonspecific automobile pictures that are captured in various realistic circumstances. The principle, procedure and simulation of the various stages of the anticipated AVLPR algorithm are designated in detail. A standardized experimentation has been prepared in MATLAB to demonstrate the fund