Basic Object Shape Detection and Tracking Using Perceptual Organization

If a robot shall learn object affordances, the task is greatly simplified if visual data is abstracted from pixel data into basic shapes or Gestalts. This paper introduces a method of processing images to abstract basic features and into higher level Gestalts. Perceptual Grouping is formulated as incremental problem to avoid grouping parameters and to obtain anytime processing characteristics. Furthermore we want to present a efficient method to track Gestalts using low-level Gestalts for motion field approximation. The proposed system allows shape detection and tracking of 3D shapes such as cubes, cones and cylinders for robot affordance learning.