A Theory of Human Stereo Vision
A Theory of Human Stereo Vision
Automated subpixel image registration of remotely sensed imagery
IBM Journal of Research and Development
A Stereo Matching Paradigm Based on the Walsh Transformation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Hi-index | 0.14 |
Results are described for an algorithm that automatically registers two stereo images. An exact area registration process which automatically removes all residual distortion due to the geometry of the stereo imaging system is demonstrated. This process allows the match judgement to be based upon an analysis of the residual differences of structure and intensity existing after all differences due to geometrical distortion have been corrected. The accuracy in general seems to be of the order of 1% of range, since the average distance to the model was about 600 mm and the RMS error in range was about 6 mm. The fact that the maximum error was greater by a factor of five indicates that there were still some mismatched points and the procedures followed in the stereo ranging algorithm were not sufficient. Further modifications to the algorithm are needed.