Desenvolvimento de um dispositivo para realizar o balanço de branco em televisores em tempo real utilizando a técnica da média de múltiplas imagens

Abstract

White balance on TVs seeks to enhance the screen display image so that it is as close as possible to the real image by removing unrealistic color projection. As these adjustments are currently performed only at the factory, this project sought to develop a low-cost device that would perform real-time white balance using the digital image processing technique called Multiple Image Average. This technique consists of obtaining an average image from N noisy images so that the resulting image gets closer and closer to a noise free image. This will be necessary since the project is based on acquiring these images through a camera. The program for processing images was developed in Python using the OpenCV computer vision library. To support the project, a survey was conducted on the color standard used in televisions, the Rec. 709, white balance, digital image processing, OpenCV, Python and Raspberry Pi, which will be the microcomputer used for real-time processing. The project is based on capturing television images with the correct and incorrect white balance, obtaining their average images without noise, and comparing them using the Pearson Correlation method, where the coefficient cut-off value was established for correctly balanced images, then a signal is sent to the TV to selfadjust, which will be simulated via an LED signaling circuit. After the tests, it was concluded that the device can assist in this procedure efficiently and reliably.