Research On Special Road Condition Recognition Algorithm In Road Traffic Video

As China’s economic development continues to improve,the development speed of road transportation is also showing signs of accelerating year by year.At the same time,road traffic violations are also increasing day by day,and the traditional road traffic monitoring system is incapable of facing this new trend.It is imperative to seek new solutions.Developed countries such as Europe and the United States unanimously use intelligent transportation systems as a powerful tool to deal with this challenge.Road traffic video recognition algorithms are the foundation and core of intelligent transportation systems.It is of great significance to carry out research in related fields.This article first designs the overall scheme of the road traffic video recognition algorithm.First,it is clear that special road conditions are speeding and congestion,and the overall design idea of first image processing and automatic identification of special road conditions is adopted to enhance the robustness of the algorithm.The algorithm is divided into three parts: video optimization processing algorithm,special road condition judgment algorithm and interactive system design.Through the preliminary investigation and system analysis,the video optimization processing is divided into weather classification and image processing for realization,and the special road condition recognition algorithm is divided into target recognition,target tracking and road condition judgment for realization.In this paper,the automatic classification processing of road traffic monitoring video is realized through the video optimization processing algorithm.First,the video automatic classification algorithm design is carried out,and the weather is divided into sunny,rainy and low-light weather.The classification feature uses a feature extraction scheme that combines local features and overall features.Dark channel features are extracted as local features,and saturated contrast features are extracted as overall features.The final features are obtained through feature fusion and a variety of machine learning solutions are used for learning.As a result,the XGBoost model was selected as the final model.Then the video automatic processing algorithm design is carried out.A variety of image processing was performed on the low-illuminance image,and a fourdimensional evaluation system was used to evaluate the processing effect.Finally,homomorphic filtering was selected as the low-illuminance image processing scheme.A linear model of the rain scene image is established for the rain image,and the operation of removing the rain pattern of the image is successfully realized through this model.In the special road condition judgment algorithm part,this paper realizes the special road condition judgment based on deep learning.The whole judgment process is divided into three parts: vehicle identification,vehicle tracking and special road condition judgment.Vehicle recognition implements a target recognition system based on the YOLOv5 framework,with a detection and recognition rate of over 85%.Vehicle tracking implements a target tracking algorithm based on the Deep Sort framework,and the success rate of complete target tracking within 2S exceeds 90%.Using the results of target recognition and tracking,an algorithm for determining speeding and congested road conditions is proposed.After testing,the algorithm can realize autonomous judgment of speeding and congested road conditions.In the last part of this article,the human-computer interaction system design and algorithm integration are realized,and the corresponding algorithm function tests are carried out.Using the integrated interactive system algorithm to achieve visual video reading,running and automatic calibration.And use the integration results to carry out related tests.The test results show that the road traffic recognition algorithm can effectively identify and judge road traffic conditions,and the congestion alarm accuracy rate reaches more than 95%.