ABSTRACT

License plate detection and recognition has become of significant interest since the
number of vehicles has increased substantially. "Ihcrefore, the need to perform.
automatic toll collection, tramc law enforcement, parking lot access control, etc. has
also been increased. proposed scheme discusses a license plate detection and
extraction method which comprises three steps: l) Felzenszwalb's Histogram of
Orientated Gradient method to extract the license plate's features and feeding those
to train Structural Support Vector Machine for training detector, 2) applying trained
detector for detection of license plate from a scene, and 3) extraction of the
text/numbers from the detected region using image enhancement techniques and
Tesseract-OCR. This proposed scheme results in a reliable technique having higher
accuracy with reasonable computational requilernent.