|
Acadêmico(a): Daniel dos Santos |
Título: Sistema Óptico para Identificação de Veículos em Estradas |
|
Abstract: |
This work presents a monitoring system for counting of vehicles, using computer vision.
Studies show that people working on monitoring systems have difficulties with attention,
which helps in the use of computational vision in this area. To verify the feasibility of using
algorithms adaptive Foreground/Background segmentation, the system compares the
algorithm NHD with the algorithm background subtraction. The software uses Fourier
descriptors in order to describe the edge of products shapes, amonget other image processing
techniques such as segmentation. For the interpretation of the representations of the vehicles,
a neural network of the type Multilayer Perceptron was implement. The combination of these
techniques made possible the development of a software for counting of vehicles, regardless
of background and brightness. |
|
|