Improved ALPR system based on Smart License Plate Character Detection Algorithm

Abstract: This paper presents improved Automated License Plate Recognition (ALPR) system based on Smart License Plate Character Detection algorithm, this system is capable of distinguishing license plates under various conditions, such as distance from the camera, rotation angle between camera and vehicle and poor illumination condition (different weather condition, different lighting condition and physical tilted or damage of license plate). In our method, improved ALPR system has three main steps: 1) Image Enhancement, 2) Character Segmentation (Smart License Plate Character Detection), 3) Character Recognition. While in regular ALPR system there are four main steps: 1) Image Enhancement, 2) License Plate Location Extraction 3) Character Segmentation, 4) Character Recognition. In image enhancement introduce Multi-Scale Adaptive NICK thresholding method [1] to achieve all objects as character candidates in the binary image; otherwise we might lose license plate characters in an input image due to complex backgrounds. Single Pass Connected Component Labelling (CCL), put the label on detected objects then new design algorithm called �Smart License Plate Character Detection� finds the license plate characters regardless of location, shape and background colour of license plate. So we eliminate the license plate location extraction as complex and time consuming individual stage [16]. All the steps simulated in Matlab 2014A software and results show the accuracy and reliability of this proposed method.

Keywords: ALPR system, Connected Components Labelling (CCL), Character Segmentation, Character Recognition, Multi-Scale Adaptive NICK thresholding Method, Smart License Plate Character Detection.