基于模板匹配方法的不均匀照度车牌图像识别

Uneven Illumination License Plate Image Recognition Base on Template Matching Method

为了快速准确识别不均匀照度车牌图像,提出一种基于改进的顶帽重构和模板匹配方法车牌识别算法。首先,使用灰度级顶帽重构进行预处理,增强图像的细节信息,减弱环境和光照等条件变化对车牌字符信息的干扰。然后,使用改进的投影分割法对二值车牌图像进行分割,提取字符图像。最后,对不同地点、不同自然条件下采集的图像进行测试,测试结果证明该算法性能优异。在此基础上,设计了一种基于字符图像全局重合度的模板匹配函数,对字符样本图片进行测试,测试结果证明该匹配函数准确可行且性能良好。

For fast and accurate recognition of uneven illumination license plate images,a new license plate recognition algorithm based on improved top-hat reconstruction and template matching was proposed. Firstly,the gray scale top-hat reconstruction was applied to do pre-processing,enhance the detail information of the image as well as weaken the interference of the environment and light conditions to the license plate character information. Secondly,the improved projection segmentation method was used to segment the binary license plate image and extract the character image. Finally,tests on the images collected from different places and different natural conditions were carried out. The test results show that the proposed algorithm has excellent performance. On the basis of the above,a template matching function based on global coincidence degree of character image was proposed,and tests were done on images of different kinds of samples. The test results show that the matching function is proved to be accurate and feasible for license plate recognition,and has good performance.