Practical alternative to estimate flow and a vehicle speed on an avenue, using Hough Transform and Morphological Operations

Abstract

In this article an estimation algorithm of vehicular flow and speed is developed, in a video AVI format for a point of Via Expresa Paseo de la República Avenue from the city of Lima, Perú. As part of digital spatial processing, Hough Transform 500x500 matrix for counted was used for the segmentation stage of the road lanes, on each of the processed frames of the AVI format video. Then, the image was binarized and morphological operators were used to label and to represent each vehicle in group of white pixeles, which facilitated the obtaining of the mass center and later compared with a reference line. This allowed the isolation of the vehicles. Also, the process was developed in the MATLAB GUI software, and a total of 30 diurnal AVI format videos each one with 8 minutes of duration were used. Regarding the percentage of success of the vehicular Flow was 95%, while the speed average estimate did not exceed 80 Km/h. In addition, the number of frames processed per second was equal to 2, which allowed a capture every 675 millisecond, to process sequences of images of 25 frames per second.