雾霾天气下的车牌识别

在雾霾天气下，图像采集设备拍摄的图片存在一系列问题，如饱和度低、细节失真、画质模糊等。文中探索了雾霾天气下的车牌识别算法，按照图像去雾、车牌定位、字符分割与识别等步骤来解决雾霾天气下传统车牌识别系统效率低、鲁棒性差等问题。该算法采用暗通道去雾，经去雾算法处理后，图像对比度、信息梯度和信息熵均得到提升；选择数学形态和边缘检测定位车牌的准确位置；利用仿射变换矫正车牌区域，结合投影法分割字符，最后使用基于支持向量机模式的识别算法来识别字符。经过处理后，车牌识别能达到较高的准确率。

License Plate Recognition in Hazy Weather

In hazy weather, there are a series of problems in the pictures taken by the image acquisition equipment, such as low saturation, distortion of details, blurred image quality, etc. This paper explores the license plate recognition algorithm in hazy weather. According to the steps of image dehazing, license plate positioning and character segmentation and recognition, the traditional license plate recognition system in hazy weather is solved. After the dehazing algorithm is processed, the image contrast, information layer and information entropy are improved. Choose the mathematical form and edge detection to locate the exact position of the license plate. Use affine transformation to correct the license plate area, combine the projection method to segment the characters, and finally use the recognition algorithm based on the support vector machine pattern to identify the characters. After processing, the license plate recognition can achieve high accuracy.