Object detection by grouping partial perception changes

Abstract

This paper deals with exhibiting and classifying objects along an image sequence. Motion is considered through the image sequence, leading to a frame by frame analysis. We first review a set of methods to detect mobile objects, knowing that our intended perception model is to include colour and get robust to flickered image acquisition. Objects are being detected due to the visual variances of observer's perceptions : different segments of the image string  are compared. The core of the image-analysis process deals with defining methods to reveal object structures. These methods piece together tasks of detecting new objects, grouping them, identifying the objects distribution along the scene. All contribute to eliminating collateral information.