An Improved Method for Vietnam License Plate Location based on Mathematic Morphology and Measuring Properties of Image Regions

Abstract

In the Automatic License Plate Recognition (ALPR) system, the License Plate Location (LPL) is the key step before the final recognition. This paper proposed an improved LPL algorithm for Vietnam vehicle license plates (LP). The proposed algorithm consists of three main modules: Pre-processing (convert RGB image to grayscale image, adjust grayscale image intensity, image binarization use Otsu method), LP candidates location (morphology opening to remove noises & dilation operation, measure properties of image regions to find candidates), LP exactly location (finding the LP angle & rotating LP, cut exactly LP region). We implemented test for 350 Vietnam vehicle images, which obtained from the actual system, the average rate of accuracy of our method is 98.64%, our results are more exactly presented methods.

Research on Characters Segmentation in One-Row and Two-Row of Vietnam License Plates

“…In this paper, we considered for statistical parameters of Vietnam LP and types of Vietnam LP (one-row LP and two-row LP), and then give out an improved method for one-row & two-row of Vietnam license plates (LP) characters segmentation based on the peak-to-valley method & statistical parameters of characters and digits, which combined pre-processing to improved image quality and remove noise. We have tested a lot of Vietnam LP, which obtained from results of our last research work [7] with different background, such us illumination, angles, sizes and types of LP, light conditions in Vietnam environment.…”