A Color and Texture Feature Based Approach to License Plate Location

Abstract

A novel license plate locating approach based on the color and texture features is presented. Firstly, the input image is converted to the hue-saturation-intensity (HSI) color space. Then a target image is obtained by applying a sequence of image processing techniques to the hue and saturation component images. After that, the space-pixel histogram of the target image is analyzed and mathematically modeled, so that the horizontal candidate is extracted. Finally, discrete wavelet transform is performed on the candidate, and the sum of the first order difference of the DWT subimages highlights the texture information of the LP area, telling the precise position of the license plate. The proposed algorithm focuses on combining the color features with the texture features, improving the locating reliability. Experiment was conducted on a database of 332 images taken from various illumination situations. The license plate detecting rate of success is as high as 96.4%.