A CNN-Based License Plate Recognition Using TensorFlow and PySpark

A CNN-Based License Plate Recognition Using TensorFlow and PySpark

Lavanya K, Bharathi K., Preethi Christina A., Satyam Chaurasia
Copyright: © 2023 |Pages: 12
DOI: 10.4018/978-1-6684-4246-3.ch001
OnDemand:
(Individual Chapters)
Available
$37.50
Current Special Offers
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

The use daily of vehicles is rising exponentially and as a result there is an increase in crimes associated with it. Many vehicles are violating the rules of traffic and so an abnormal number of accidents occur leading to a rise in the crime rates linearly. In order for any vehicle to be recognized, its license plate number is needed. Therefore, the vehicle license plate detection plays a notable role. The optical character recognition (OCR) is one effective way to scan number plates and recognize the text found in the digital image, containing the license plate number into machine readable text which can then be used to track the vehicles. The image of the number plate is first captured, processed, and every character present in the number plate is read for perfect recognition. The optical character recognition model is trained using TensorFlow. Spark's in-memory data engine can perform tasks rapidly in multi-stage jobs. Therefore, TensorFlow and Spark are used together to train and apply the OCR model to perform the license plate recognition swiftly.
Chapter Preview
Top

Literature Survey

In the pre-processing of images with the OCR software, data sets containing non intersecting images are subjected to a neural network. Images and the data required for training are collected, the attributes are detected, then the accuracy of detection is improved. The OCR is then optimised to find different ways in which optimal results can be obtained with fewer failures. (Madhumitha and Dhivya, 2020)

The improved segmentation method involves selecting the image and removing noise. The interested area of image is detected, then the location of the license plate is obtained using edge detection (Heng Sun et al.,2018). Each character in the image undergoes segmentation individually. The template matching method is then used with the help of correlation for recognition of characters in the number plate. (Balaji and Rajesh, 2017)

Complete Chapter List

Search this Book:
Reset