On Computing the Next Look Camera Parameters in Active Vision