Color and Shape Recognition

Abstract

The object "car" and "cat" can be easily distinguished by humans, but how these labels are assigned? Grouping these images is easy for a person into different categories, but its very tedious for a computer. Hence, an object recognition system finds objects in the real world from an image. Object recognition algorithms rely on matching, learning or pattern recognition algorithms using appearance-based or feature-based techniques. In this thesis, the use of color and shape attributes as an explicit color and shape representation respectively for object detection is proposed. Color attributes are dense, computationally effective, and when joined with old-fashioned shape features provide pleasing results for object detection. The procedure of shape detection is actually a natural extension of the job of edge detection at the pixel level to the difficulty of global contour detection. A tool for a systematic analysis of edge based shape detection is provided by this filtering scheme. This enables us to find distinctions between objects based on color and shape.