A Black Car Recognition Method Based on Trajectory Features and Person-vehicle Correlation

Abstract

The invention discloses a black car identification method based on track features and human-vehicle association, and specifically relates to the field of intelligent transportation technology. The specific steps are as follows: Step 1: Confirm whether all registered vehicles have operating qualifications; Step 2: Determine the list of alternative features; Step 3: Calculate the relevant information of people and vehicles; Step 4: Screen the list of salient features of commercial vehicles; Step 5: Determine the black car target. Through the setting of steps 1 to 2, the present invention uses the characteristics of legally operating vehicles and information related to people and vehicles to find vehicles with similar characteristics to find vehicles that have no operating qualifications but may be in operation. Through steps 3 to The setting of step 5 uses calculations to screen the characteristics of black cars. The characteristics of black cars are clearly described, accurately defined, and the recognition accuracy is high, which is conducive to maintaining the normal order of the transportation market, reducing the occurrence of traffic accidents, and protecting the legality of passengers. Rights and travel safety are conducive to social stability.

Description

technical field [0001] The invention relates to the technical field of intelligent transportation, in particular to a black car identification method based on track features and human-vehicle association. Background technique [0002] Black cars refer to vehicles that have not gone through any relevant procedures in the transportation management department, obtained operating licenses, and are operating illegally with paid services. In recent years, the number of black cars has increased year by year, which has seriously disrupted the normal order of the transportation market, increased the occurrence of traffic accidents, damaged the legitimate rights and interests of passengers, seriously affected the travel safety of passengers, and brought unstable factors to the society. . [0003] Therefore, government departments are also paying more and more attention to the work of cracking down on black cars, and have carried out a series of research on black car identification. E...

AI Technical Summary

Problems solved by technology

In recent years, the number of black cars has increased year by year, which has seriously disrupted the normal order of the transportation market, increased the occurrence of traffic accidents, damaged the legitimate rights and interests of passengers, seriously affected the travel safety of passengers, and brought unstable factors to the society.

[0003] Therefore, government departments are also paying more and more attention to the work of cracking down on black cars, and have carried out a series of research on black car identification. Existing black car identification is mostly identified through authority authentication, machine learning algorithms, k-means algorithm hard aggregation, and manual methods. It is obviously inappropriate to use the existing black and white car samples for machine learning classification algorithms; in addition, using the k-means algorithm for hard clustering will also have the problem of uncontrollable clustering results, because the black cars are small It is very likely that the black car samples of the clustering results are diluted in different classes, which leads to the need to re-model the problem. Even if the existing feature system is adjusted, the expected results may not necessarily be expected. At the same time, because the clustering results are in The business explanation ability is poor, so the final model is not convincing; and these methods cannot monitor the increase or decrease of black cars and the travel route in real time, and cannot complete the timely update of information, which is not conducive to the tracking of black cars, so it is necessary Putting effort into feature screening, but the traditional emphasis on feature engineering is often not enough in practice, the definition of black cars is too loose, and the feature description is not clear, resulting in inaccurate data analysis and data mining

Images