DEFF Research Database (Denmark)

In this work, we address the problem of road interpretation for driver assistance based on an early cognitive vision system. The structure of a road and the relevant traf?c are interpreted in terms of ego-motion estimation of the car, independently moving objects on the road, lane markers and large scale maps of the road. We make use of temporal and spatial disambiguation mechanisms to increase the reliability of visually extracted 2D and 3D information. This information is then used to interpret the layout of the road by using lane markers that are detected via Bayesian reasoning. We also estimate the ego-motion of the car which is used to create large scale maps of the road and also to detect independently moving objects. Sample results for the presented algorithms are shown on a stereo image sequence, that has been collected from a structured road.

CERN Document Server

This paper describes a new method, HGV2C, for pattern analysis. The HGV2C method involves the construction of a computer ego (CE) based on an individual object that can be either a part of the system under analysis or a newly created object based on a certain hypothesis. The CE provides a capability to analyze data from a specific standpoint, e.g. from a viewpoint of a certain object. The CE is constructed from two identical copies of a query object, and its functioning mechanism involves: a hypothesis-parameter (HP) and infothyristor (IT). HP is a parameter that is introduced into an existing set of parameters. The HP value for one of the clones of a query object is set to equal 1, whereas for another clone it is greater than 1. The IT is based on the previously described algorithm of iterative averaging and performs three functions: 1) computation of a similarity matrix for the group of three objects including two clones of a query object and a target object; 2) ...