An enhanced vehicle licence plate recognition system based on neural networks

Automatic license plate recognition (ALPR) is an important forensic functionality
for law enforcement and cyber forensics departments for analysing
cctv videos. In many cases number plate image may be blurred or may be of
low resolution due to over speed of vehicles, under adverse atmospheric conditions
or due to low resolution cameras. In this scenario, super-resolution
can be an excellent solution to overcome such limitations. Consecutive frames
in a video may contain dierent information that could be integrated into a
single image, richer in details. Data set required for the project is acquired
through cctv footages and trac monitoring videos. The trained software
model should able to enhance the sequence of frames containing license plates
in low quality real world trac videos captured by closed circuit television
cameras. The ultimate aim of the project is to enhance and recognize low
quality or low resolution blurred images of licence plates from cctv trac
video footages. Image enhancement is implementing using super resolution
of multi frames from videos and recognition part is implementing using optical
character recognition. The scope of project is mainly intended for the
use of law enforcements for detecting the non recognizable images of licence
plates of suspected vehicles.

MSC CS 2017-2019 INT Pradeep Kumar K