A Pixel Dissimilarity Measure That Is Insensitive to Image Sampling

Because of image sampling, traditional measures of pixel dissimilarity can assign a large value to two corresponding pixels in a stereo pair, even in the absence of noise and other degrading effects. We propose a measure of dissimilarity that is provably insensitive to sampling because it uses the linearly interpolated intensity functions surrounding the pixels. Experiments on real images show that our measure alleviates the problem of sampling with little additional computational overhead.

[1]  Peter N. Belhumeur,et al.  A binocular stereo algorithm for reconstructing sloping, creased, and broken surfaces in the presence of half-occlusion , 1993, 1993 (4th) International Conference on Computer Vision.

[2]  H. K. Nishihara,et al.  Practical Real-Time Imaging Stereo Matcher , 1984 .

[3]  Daniel Scharstein,et al.  Matching images by comparing their gradient fields , 1994, Proceedings of 12th International Conference on Pattern Recognition.

[4]  Ramin Zabih,et al.  Non-parametric Local Transforms for Computing Visual Correspondence , 1994, ECCV.

[5]  Ingemar J. Cox,et al.  A Maximum Likelihood Stereo Algorithm , 1996, Comput. Vis. Image Underst..

[6]  Alain Crouzil,et al.  A new correlation criterion based on gradient fields similarity , 1996, Proceedings of 13th International Conference on Pattern Recognition.

[7]  David Mumford,et al.  A Bayesian treatment of the stereo correspondence problem using half-occluded regions , 1992, Proceedings 1992 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[8]  Peter Seitz,et al.  Using Local Orientational Information As Image Primitive For Robust Object Recognition , 1989, Other Conferences.

[9]  Aaron F. Bobick,et al.  Disparity-Space Images and Large Occlusion Stereo , 1994, ECCV.