A Multiple-baseline Stereo Method
Takco Kanade, Masatoshi Okutomit, Tomoharu Nakahara · 2002
This paper presents a stereo matching method which uses multiple stereo pairs with various baselines to obtain precise distance estimates without suffering from ambiguity. In stereo ptvcessing. a shon baseline means that the estimated distance wi l l be less precise due to n a m w triangulation. For more precise d ismce estimation. a longer baseline is desired Wth a longer baseline. however. a larger dispariv range must be searched ro find a march. As a result, marching is more di’cult and there is a greater possibiliry of a false match. So there is a trade-off beween precision and accuracy in matching. The stereo matching method presented in this paper uses multiple stereo pairs with different baselines generated by a lateral displacement of a camera Matching is performed simply bv computing the sum of squared-digerencc (SSD) values. The SSD functions for individual stereo pairs are represented with respect to the inverse distance (rather than the disparity as is usually done). and then are simply added 10 produce rhe sum of SSDs. This resulting function is called the SSSD-in-inrerse-distance. We show that the SSSD-in-inverse-distance~ction uhibiu a unique and clear minimum at the correct matching position even when the underlying inrensily patterns of the scene include ambiguities or repeiitive patterns. An advantage of this method is that we can eliminate false marches and increase precision without anv search or sequential filtering. This paper first defies a stereo algorithm based on the SSSDin-inverse-distance and presents a marhemarical analysis 10 show how the algorithm can remove ambiguip and increusr precision. Then, a few experimental results with real siereo images are presenred to demonstrate the effectiveness of the algorithm