一、车牌识别技术原理

1、车牌识别系统框架 真地车牌识别系统采用高度模块化的设计，真地将车牌识别过程的各个环节各自作为一个独立的模块，真地采用国际领先的计算机智能算法技术，首先通过摄像头提取车牌视频图像，对获取的每一帧图像，真地利用最新的高效视频检测技术对车牌进行定位和跟踪，车牌识别摄取从中自动提取车牌图像，然后经过车牌精确定位、切分和识别模块准确地自动分割和识别字符，得到车牌的全部字符信息以及颜色信息。系统框架如图所示： The real license plate recognition system uses a highly modular design. It really takes each link of the license plate recognition process as an independent module. It really adopts the international leading computer intelligent algorithm technology. First, it extracts the license plate video image through the camera. For each frame of image, it really uses the latest high efficiency. Video detection technology locates and tracks the license plate, and the license plate image is automatically extracted from the license plate recognition. Then, the characters are automatically segmented and recognized accurately by the accurate license plate location, segmentation and recognition module, and the whole character information and color information of the license plate are obtained. The system frame is shown in the figure: 2、车牌识别车牌定位模块 整个停车场智能系统车牌定位模块是一个十分重要的环节，是后续环节的基础，其准确性对整体系统性能的影响巨大。本系统实现了一种完全基于学习的多种特征融合的车牌定位新算法，适用于各种复杂的背景环境和不同的摄像角度。真地车牌识别系统同步内置的区域设置功能，可以很大程度上减少外部环境对于车牌识别的干扰，达到最好的识别效率。 The license plate positioning module of the whole parking intelligent system is a very important link, is the basis of the follow-up link, its accuracy has a great impact on the overall system performance. This system implements a new license plate location algorithm based on learning and multi-feature fusion, which is suitable for various complex background environments and different camera angles. The real license plate recognition system synchronizes the built-in area setting function, can greatly reduce the external environment for the license plate recognition interference, to achieve the best recognition efficiency. 3、车牌矫正及定位模块 车牌识别受拍摄条件的限制，图像中的车牌总不可避免的存在一定的倾斜，车牌识别需要一个矫正和定位环节来进一步提高车牌图像的质量，为切分和识别模块做准备。真地车牌识别系统使用独创的精心设计的快速图像处理滤波器，计算速度快，利用车牌整体信息，避免了局部噪声带来的影响，另一个优点就是通过对多个中间结果的分析还可以对车牌进行定位，进一步减小非车牌区域的影响。真地车牌识别系统较之其他公司的车牌识别系统，对于倾斜车牌图片的支持要更多，一般的车牌识别软件，支持的倾斜角度是±5°，该车牌识别软件，能支持±45°的倾斜角度，并且通过内部倾斜校正算法，不会因为倾斜角度的增加而降低识别率。 The license plate recognition is limited by the shooting conditions, the license plate in the image always inevitably exists a certain tilt, the license plate recognition needs a correction and positioning link to further improve the quality of the license plate image, in order to prepare for segmentation and recognition module. True Land License Plate Recognition System uses original and elaborately designed fast image processing filter, which has fast calculation speed and avoids the influence of local noise by utilizing the overall information of the license plate. Another advantage is that it can locate the license plate through the analysis of many intermediate results and further reduce the influence of non-license plate area. . True Land License Plate Recognition System has more support for tilted license plate pictures than other license plate recognition systems. General license plate recognition software supports tilt angle of +5 degrees. This license plate recognition software can support tilt angle of +45 degrees, and through internal tilt correction algorithm, it will not increase tilt angle. Increase and decrease the recognition rate. 4、车牌切分模块 车牌切分模块利用车牌文字的灰度、颜色、边缘分布等各种特征，能较好地抑制车牌周围其他噪声的影响，并能容忍一定倾斜角度的车牌。 The license plate segmentation module makes use of the gray level, color, edge distribution and other features of the text of the license plate, which can better suppress the influence of other noise around the license plate, and can tolerate a certain tilt angle of the license plate. 5、车牌识别模块 在车牌识别模块中，采用了多种识别模型相结合的方法，构建了一种层次化的字符识别流程，有效地提高了字符识别的正确率。另一方面，在字符识别之前，使用计算机智能算法对字符图像进行了前期处理，不仅保留了图像信息，而且提高了图像质量和相似字符的可区分性，保证了字符识别的可靠性准确性。 In the license plate recognition module, a hierarchical character recognition process is constructed by combining several recognition models, which effectively improves the accuracy of character recognition. On the other hand, before character recognition, the computer intelligent algorithm is used to pre-process the character image, which not only retains the image information, but also improves the image quality and distinguishability of similar characters, ensuring the reliability and accuracy of character recognition. 真地车牌识别行业领导者，专业研发停车场系统10年以上，一直以创新稳定的技术，服务于全球各大停车场，使通行更安全，方便快捷。