Real-time correllation - based stereo vision with reduced error borders

The research reported here describes an approach to real-time stereo vision using non specialised computer equipment. In particular the new methods improve standard correlation considerably. The extensive results presented in terms of performance allow us to use simple cameras to perform excellent image analysis. This work has been used and extended by one of the authors in real applications for mobile robots. The paper appeared in one of the leading journals in Computer Vision with an impact factor of 6.085