加油站复杂环境下车牌定位算法研究

Research on the Algorithm of License Plate Localization in Gas Station with Complicated Environment

车牌定位技术是整个车牌识别系统中最关键的一步。车牌定位的准确程度决定了后期车牌字符分割和字符识别的效果。现在已有的车牌定位算法大多数适用于路面情况简单、背景环境单一的情况,而对于加油站等复杂环境适用性不高。本设计提出加油站复杂环境下的改进的车牌定位算法,该算法通过边缘检测来初步定位车牌,并通过迭代融合算法,多次的筛选候选区域。最后对车牌进行阈值矫正,从而达到车牌精确定位效果。实验结果表明本设计中的定位算法,其准确率与时效性明显的高于传统的定位算法。该算法解决了传统定位算法只能应用于简单路面环境的局限性的问题,适合于车流量大、车况复杂的加油站现场,可行性高,定位准确,具有一定的实用价值。

License plate locating technology is the key point of the whole license plate recognition system. The extent of locating accuracy can largely determine the later character segmentation and recognition effect. Most of the existing license plate localization al- gorithms are suitable for simple conditions, but are not suitable for the complex environment such as gas stations. This paper gives a design of an improved license plate location method for gas station in complex environment. It uses an algorithm based on edge detec- tion to preliminary locate the plate, and an iterative fusion algorithm is used to repeatedly screen the candidate region. Finally, the threshold ＇correction of license plate helps to achieve the effect of accurately locating the license plate. The experimental results show that the accuracy and timeliness of the designed localization algorithm in this article is significantly higher than the traditional localiza- tion algorithm, and the algorithm can solve the problem that the traditional localization can only be limited to simple road environment. The algorithm is suitable for busy traffic and complex site of the gas station site, which has strong practicability and accurate positio- ning, and thus has certain practical value.