论文部分内容阅读
The number of vehicles is increasing exponentially day by day all over the world. A few decades ago, in a span of 10 years (1990–1999), 39.2 million cars were sold worldwide in comparison to 81.5 million solely in 2018 [1]. The growing number of vehicles causes the increasing complexity of transportation infrastructure and congestion as well as car-related crimes. For instance, in 1982 in Chicago, the total delayed amount of time per commuter for one year was 31 hours, while in 2018, it became 61 hours (97% growth) [2]. As for the car crimes, around 25%of traffic and parking tickets that were issued in Washington, US last year went unpaid. That’s around $85 million net worth and the tendency shows that the crime rate is increasing and not only in US [3]. Moreover, a huge number of drivers all over the globe have their license plates stolen every year. Therefore, spreading a real-time information to commuters and drivers is necessary.
Recently, the availability of GPS, camera on smartphones and abundance of license plate recognition algorithms made a mobile phone-based traffic monitoring system highly convenient. A person owning a smartphone with a traffic app is aware of cars movement, on-road accidents and vehicle-related crimes. In addition, some applications are designed in a way that users can contribute and share their collected data with other clients. One of the most popular platforms for a smartphone is Android. There are more than 2 billion active devices using Android operating system (OS) [4]. Therefore, building an Android-based traffic monitoring system can help many users struggling on the road.
In this research work, we will introduce a transportation monitoring system for Lithuania. A vehicle-dense country, lacking a reliable traffic application. The android based application is community-driven as the data is collected by its users. The LTTMS implements OpenALPR Android library to do the license number recognition. For higher identification accuracy (~90%) of the plate, the optical character recognition (OCR) engine, called Tesseract, was trained according to the format of Lithuanian plate number. The modifications of OCR were made utilizing 221 different images of Lithuanian license plate as a training data. In addition, Google Maps SDK was implemented into LTTMS to track the location of every single traffic event that is tagged by the user. This solution enables us to generate precise coordinates and create a great visualization for the users on a general map of all traffic issues with its details. A connection between the Android client and the database via the server was established by building PHP scripts. An efficient code allowed us to develop a fast communication, which is necessary so the clients can always stay updated about the traffic issues by themselves and help others to prevent it.
Recently, the availability of GPS, camera on smartphones and abundance of license plate recognition algorithms made a mobile phone-based traffic monitoring system highly convenient. A person owning a smartphone with a traffic app is aware of cars movement, on-road accidents and vehicle-related crimes. In addition, some applications are designed in a way that users can contribute and share their collected data with other clients. One of the most popular platforms for a smartphone is Android. There are more than 2 billion active devices using Android operating system (OS) [4]. Therefore, building an Android-based traffic monitoring system can help many users struggling on the road.
In this research work, we will introduce a transportation monitoring system for Lithuania. A vehicle-dense country, lacking a reliable traffic application. The android based application is community-driven as the data is collected by its users. The LTTMS implements OpenALPR Android library to do the license number recognition. For higher identification accuracy (~90%) of the plate, the optical character recognition (OCR) engine, called Tesseract, was trained according to the format of Lithuanian plate number. The modifications of OCR were made utilizing 221 different images of Lithuanian license plate as a training data. In addition, Google Maps SDK was implemented into LTTMS to track the location of every single traffic event that is tagged by the user. This solution enables us to generate precise coordinates and create a great visualization for the users on a general map of all traffic issues with its details. A connection between the Android client and the database via the server was established by building PHP scripts. An efficient code allowed us to develop a fast communication, which is necessary so the clients can always stay updated about the traffic issues by themselves and help others to prevent it.