This dissertation refers the use of stationary cameras to track persons during human-robot cooperation experiments. To this end, a method to predict perspective-related shifts in the color distribution of pedestrians is presented. A gemoetric model of a pedestrian is employed, which allows for the estimation of color distributions of single body parts and their share in the resulting aggregate appearance. An experiment conducted to verify the proposed method, using 40 ceiling-mounted cameras, yielded a 12% reduction in average deviation of estimated position from the ground truth.
«
This dissertation refers the use of stationary cameras to track persons during human-robot cooperation experiments. To this end, a method to predict perspective-related shifts in the color distribution of pedestrians is presented. A gemoetric model of a pedestrian is employed, which allows for the estimation of color distributions of single body parts and their share in the resulting aggregate appearance. An experiment conducted to verify the proposed method, using 40 ceiling-mounted cameras, yi...
»