Distance and relative speed estimation of vehicles moving in the same direction

Intelligent vehicle systems often use expensive hardware to calculate distances and speeds of surrounding objects. This paper presents two algorithms for estimating the relative speed of vehicles using monocular cameras. The first algorithm relies on license plate detection with a known size, while a modified version incorporates car dimension calculations. The second algorithm utilizes pre-calibrated cameras based on marked driveways. An experiment was conducted to compare the efficiency of the algorithms. Experiments showed that the most accurate results for estimation the distance to vehicles were obtained by the preliminary marking method with an error close to 0%. However, this method has significant limitations, such as camera fixation and stabilization requirements. The second most accurate method, based on using vehicle width values, demonstrated an error range of 1.5% to 3.3%. Unlike the first method, this algorithm does not require strict camera fixation and stabilization.