Desarrollo de un sistema Web para el registro automático de placas vehiculares utilizando cámaras de vigilancia en parqueaderos

Abstract

The objective of this work is to develop a web system for the registration of license plates using surveillance cameras in parking lots using computer vision techniques. As part of the vehicle detection, a background subtraction algorithm has been used to obtain the movement in a specific area. Once the motion is detected, the frame is extracted, stored and processed by means of grayscale transformation and noise reduction. Once this process is completed, the Canny edge detection algorithm is applied, which identifies the 4 contours belonging to a vehicle license plate. The next step will be to apply OCR, which will return the text inscribed on the license plate. The results obtained compare the performance of the system and the time interval in which it is evaluated, whereby, it is detected that the highest point of vehicle detection occurred in the range of 10:30 to 10:45 and the detection of plates occurs in the range of 11:30 to 11:45. In conclusion, the performance of the system generates an effectiveness ranging from 70% to 88% in recordings during the day, in order to obtain natural light illumination, while excluding adverse weather conditions.