With the sustained and rapid development of the social economy and the rapid growth of urban vehicles, urban expressway has developed rapidly. As the roadmap of urban traffic, the urban expressway has a relatively high and stable driving speed, and also bears a large amount of urban traffic. However, in recent years, with the expansion of the city scale, the congestion of urban expressways has become increasingly severe. In various areas of the merged area, due to various factors such as mismatched traffic capacity and reduced driving speed, it is easy to cause deterioration of road conditions in a short period of time and cause more secondary accidents. In order to reduce the incidence of expressway traffic accidents and to avoid as much as possible the casualties and property losses caused by accidents, Intelligent Transportation Systems (ITS) supported by information technology, data communication transmission technology, control technology and traffic engineering were introduced. With the rise of artificial intelligence and the continuous development of ITS, the video acquisition methods of real-time traffic flow data and the image recognition ability of video sequences continue to improve, providing theoretical basis and technical support for the research of urban expressway parallel pattern recognition. Aiming at the shortcomings of traditional pattern recognition methods, such as weak anti-interference to complex traffic environment and low correct recognition rate, this paper studies the pattern recognition method based on image processing, and selects the fuzzy C-means clustering in the currently used clustering methods (FCM) algorithm Because the FCM algorithm cannot obtain the global optimal solution and need to determine the number of cluster categories in advance, this paper uses ReliefF algorithm and Particle Swarm Optimization (PSO) to compare the feature weight and number of clusters of traditional FCM algorithm. Make improvements. Through the experimental analysis of the acquired video images, the results show that the improved FCM algorithm based on the proposed method has better real-time and accuracy in the application of urban expressway parallel pattern recognition.

随着社会经济的持续快速发展和城市车辆的快速增长，城市高速公路发展迅速。作为城市交通的路线图，城市高速公路具有相对较高且稳定的行驶速度，并且承载着大量的城市交通。然而，近年来，随着城市规模的扩大，城市高速公路的拥堵变得越来越严重。在合并区域的各个区域中，由于诸如交通容量不匹配和行驶速度降低之类的多种因素，很容易在短时间内导致道路状况的恶化并引起更多的二次事故。为了减少高速公路交通事故的发生，并尽可能避免事故造成的人员伤亡和财产损失，介绍了由信息技术，数据通信传输技术，控制技术和交通工程技术支持的智能交通系统。随着人工智能的兴起和ITS的不断发展，实时交通流数据的视频采集方法和视频序列的图像识别能力不断提高，为城市高速公路并行模式的研究提供了理论依据和技术支持。承认。针对传统模式识别方法的缺点，如对复杂交通环境的抗干扰能力弱，正确识别率低等，本文研究了基于图像处理的模式识别方法，并在目前使用的聚类方法（FCM）算法中选择模糊C均值聚类由于FCM算法无法获得全局最优解且需要提前确定聚类类别的数量，因此本文使用ReliefF算法和粒子群优化（ （PSO）比较传统FCM算法的特征权重和聚类数。进行改进。通过对采集到的视频图像进行实验分析，结果表明，基于该方法的改进型FCM算法在城市高速公路并行模式识别中具有较好的实时性和准确性。本文使用ReliefF算法和粒子群优化算法（PSO）对传统F​​CM算法的特征权重和聚类数目进行比较。进行改进。通过对采集到的视频图像进行实验分析，结果表明，基于该方法的改进型FCM算法在城市高速公路并行模式识别中具有较好的实时性和准确性。本文使用ReliefF算法和粒子群优化算法（PSO）对传统F​​CM算法的特征权重和聚类数目进行比较。进行改进。通过对采集到的视频图像进行实验分析，结果表明，基于该方法的改进型FCM算法在城市高速公路并行模式识别中具有较好的实时性和准确性。