SILVA, R. H. P.; http://lattes.cnpq.br/2335332893875661; SILVA, Roberto Higino Pereira da.
Abstract:
The images objects embedded extraction has several applicationns in theautomation area, such as: patterns recognition in surveillance systems, robots vision and others. An image statistical extraction algorithm in the RGB colors space was implemented in a DSP`plataform and the obtained results analysis are presented in this dissertation. We propose a new statistical method for objects extraction in an unknown background (non-homogenous or homogenous), using the YCbCr space. It uses maximum metric to determine its distances betwenn the space vectors, being capable to suport small global variations and places of brightness. They are presented the simulation results that validate the algorithm functionality for applications that demand performance and be tolerant to a determined error interval.