Sistema visual de segmentação de objetos robusto à iluminação para monitoramento automático de tráfego de veículos terrestres

AUTOR(ES)

Gabriel Irribarem Soares Ruas

DATA DE PUBLICAÇÃO

2009

RESUMO

This study presents a proposal for a moving object segmentation system that is resistant to changes in lighting conditions and can detect and suppress moving shadows. The technique of background subtraction using mixtures of gaussians provided a basis for this study. To better deal with illumination changes, that technique was adapted to use the HSV color space. The influence of shadows over the the background model was analyzed. Furthermore, a model to reduce their influence over the object segmentation process was developed. The effects of automatic gain control caused by illumination changes in the scene were also studied. Using the concept of scene stability, a method to model and control the effects of camera gain on the scene was proposed. Shadow supression yielded a correct rate of 36.63 %, which is similar to results obtained using the original technique proposed by Stauffer. The false positive rate was reduced from 71.38 % to 42.65 %. In addition, the proposed methodology was also able to significantly reduce the influence of sudden illumination variations on the background subtraction results.

ASSUNTO(S)

controle automático de ganho visão computacional segmentação engenharia eletrica subtração de fundo supressão de sombras

Documentos Relacionados

• Sistema automático para o controle eficiente de iluminação para multiplas lâmpadas fluorescentes
• Um sistema automatico de localização de objetos no espaço utilizando processamento paralelo
• Alfabetização visual para a produção de objetos educacionais
• Um protÃtipo de sistema para planejamento automÃtico de missÃes aÃreas.
• Monitoramento automÃtico de pH, nÃquel e cloreto em banhos de nÃquel tipo watts utilizando um sistema SIA