Using Redundant Visual Information From Different Dimensions for