Desarrollo de un sistema de control para el posicionamiento del servomecanismo de una cámara de identificación de vehículos infractores por semáforo mediante reconocimiento de imágenes utilizando visión artificial.

Abstract

The present work aimed at developing a control system for the servo device positioning of a webcam detecting red light violations with image recognition using artificial vision in the city of Riobamba and verifying the hipothesis. The investigation was experimental due to the causeeffect

realtionship. This investigation consisted of two phases: control systems and artificial vision. In the control system phase, a study was carried out to determine the dynamics of the plant system phase in a trasnsfer function using ARX (autoregressive with exogenous input) to identify the linear behavior. Then a proportional integral (PI) controller was developed adapting to the system for its control. A desing was done for a slow control with values of Kp (proportional gain) = 0.1, Ki = 0.3 and N = 0.8 and for a fast control with values of Kp = 1, Ki = 0 and N = 1E-9. In the second phase, an algorithm was developed in Matlab to process images aiming at identifying the red light using filters. This stop sign made the servo device controller move according to set point. The controller test and results were carried out during day, afternoon and evening using a fast and slow control for each one. Using the simple random samples showed an efficiency of 100% during day and afternoon with a fast and slow control. For the fast controller test during evening with a sample of 392, failures of 1% and success of 99% were gotten. For the slow controller test during evening with a sample of 392, the same failures and successes were gotten. It is concluded that artificial visión may be used for the servo device position of a webcam capable of reacting to the red light activation to identify violations using a vision camera and processing images on computer. It is recommended to use a sensor of angular displacement calibrated and adapted to the device to read the positioning error appropriately.