Raw data processing in Bayesian vehicle tracking algorithms

The article addresses one of the significant issues related to the use of vehicle tracking techniques in Intelligent Transportation Systems, which is the preparation of input data for the tracking algorithm. Due to the fact that typical tracking algorithms are not capable of correctly determining motion trajectories for objects represented by signals with a level close to noise, Track-Before-Detect (TBD) technology is used for this purpose. Such algorithms are based on a Bayesian approach using full information about the object. The article illustrates the tracking effects that can be achieved through the use of raw data, which allows for effective tracking of objects located close to each other that are characterized by different colors. An exception is made for objects with a color similar to the background, which is estimated and eliminated.PROCESSING OF RAW DATA IN BAYESIAN VEHICLE TRACKING ALGORITHMS
The paper is related to an important issue related to the application of vehicles tracking technology for the Intelligent Transportation System, which is the preprocessing of the input data for the tracking algorithm. Since typical tracking algorithms are unable to detect the motion trajectories properly for the objects represented by the signals when their level is similar to the noise, the Track-Before-Detect (TBD) approach is used in such cases. Such algorithms are based on the Bayesian approach utilising full information about the object. In the paper some results of tracking are demonstrated, which can be obtained due to the usage of the raw data, allowing an effective tracking of objects located close to each other, which are characterised by different colours. An exception is related to the objects with similar colour to the background, which is estimated and eliminated. (…)

Raw data processing in Bayesian vehicle tracking algorithms – PDF file for download