Detecção de fluxo de pessoas utilizando subtração de imagens

Abstract

This work presents the development of a program that detects the flow of people in a sequence of images. The program uses the C++ programming language and OpenCV, which is a multiplatform library aimed at developing applications in the area of Computer Vision. Digital image processing techniques related to motion detection in images, such as image subtraction and image thresholding, were explored. After applying these techniques, counting people is done from the centroid of each detected moving object. Finally, the flow is shown in the result image, that is, the number of people in one direction subtracted from the number of people in the other direction, and the number of analyzed images.