Grid-based real-time image processing (GRIP) algorithm for heterogeneous traffic

The paper presents a fast algorithm for real-time image processing for counting and classification of vehicles in heterogeneous traffic recorded using a single stationary camera.

The proposed method uses a single feature as the base parameter which is given by the user to classify the vehicles into four different classes. The algorithm has an error of 6.1% on an average for the total count when studied under varying illumination and weather conditions.