Implementação de um detector de movimentos à partir de streamings de vídeo

Abstract

The study focused on the special case of detecting movement in an environment. This environment was being recorded by a fixed camera. For data acquisition, a video streaming was used, captured through a camera with USB connection. After capturing the data, they were decomposed into individual frames and treated as single photographs. The so-called background removal technique was applied to each of these images. Frequency and morphological convolutional filters were also used. They played a cleaning and highlighting role in target regions of the image. Combined, it was possible to highlight the region of the original image where the movements occurred. Whenever movement was detected, a text event was generated informing the number of the photograph. Its correct functioning has been completed. The case of its possible extension of applicability if integrated with other systems was also mentioned. In this case, both storage and transmission of data have the potential to be reduced. Thus, when implemented, a passive economy of resources will be possible. Finally, tests with other hardware devices were suggested to improve their energy performance.