A Novel and Robust Algorithm for License Plate Location Using Perceptual Salient Features

作者：Yong, Wang; Guo You, Wang; Ran, Wang; Yuan Chun, Xia; Zhong, Chen

来源：Applied Mechanics and Materials, 2014, 513-517: 1414-1420.

DOI：10.4028/www.scientific.net/amm.513-517.1414

<jats:p>License plate location (LPL) plays an important role in license plate (LP) recognition systems. Most existing methods can only deal with LPL well in fixed environment such as stable illumination, simple background or constant distance between cameras and vehicles. And the successful rate of LPL is quite affected by those non-LP objects in the background that share similar features with license plate regions. To overcome these problems, in this paper, we propose a novel LPL method using perceptual salient features, which is based on the visual attention mechanism to choose the most salient features for license plate. The proposed approach consists of two main steps. Firstly, candidate license plate regions consistent with the visual perception are detected based on perceptual salient features from the saliency map. Then, the candidate license plate regions are sifted to distinguish a license plate from background, using salient features which are selected and organized by minimizing probability of error. The proposed algorithm was tested with 1942 real images captured in various conditions and achieved an average accuracy of 98.51%. The experiment result shows that our algorithm is robust to the interference of non-LP objects as well as the change of illumination, view angle, position, size and color of license plates.</jats:p>

• 出版日期2014-2
• 单位华中科技大学