一个车牌识别系统是否实用，最重要的指标是识别率。国际交通技术作过专门的识别率指标论述，要求是24小时全天候全牌正确识别率85%～95%。信路通的车牌识别系统在实际应用中已经达到了全牌正确识别率90%以上。 为了测试一个车牌识别系统识别率，需要将该系统安装在一个实际应用环境中，全天候运行24小时以上，采集至少1000辆自然车流通行时的车牌照进行识别，并且需要将车辆牌照图像和识别结果存储下来，以便调取查看。然后，还需要得到实际通过的车辆图像以及正确的人工识别结果。之后便可以统计出以下识别率：
Whether a license plate recognition system is practical, the most important index is the recognition rate. International traffic technology has made a special discussion on the recognition rate index, and the requirement is that the correct recognition rate of the 24-hour all-weather and all-weather license plate is 85% - 95%. The license plate recognition system of xinlutong has achieved a correct recognition rate of more than 90% in practical application. In order to test the recognition rate of a license plate recognition system, it is necessary to install the system in a practical application environment, run for more than 24 hours all day long, collect at least 1000 vehicle license plates when natural traffic flows through for recognition, and store the license plate image and recognition results for easy retrieval and viewing. Then, we also need to get the actual vehicle images and the correct manual recognition results. After that, the following recognition rates can be calculated:
1、自然交通流量的识别率=全JDG金属穿线管 不锈钢链板 隔离墩模具 检查井钢模具 流水槽模具 阶梯护坡模具 护坡砖模具 遮板模具牌正确识别总数/实际通过的车辆总数
1. Recognition rate of natural traffic flow = total number of correct identification of full license plate / total number of vehicles actually passing through
2、可识别车牌照的百分率=人工正确读取的车牌照总数/实际通过的车辆总数
2. The percentage of recognizable license plates = the total number of manually read license plates / the total number of vehicles actually passed
3、可识别全牌正确识别率=全牌正确识别的车牌照总数/人工读取的车牌照总数这三个指标决定了车牌识别系统的识别率，诸如可信度、误识率等都是车牌识别过程中的中间结果。
3. The three indexes determine the recognition rate of the license plate recognition system, such as the credibility and the false recognition rate, which are the intermediate results in the process of license plate recognition.