Rohith MV, Gowri Somanath, Chandra Kambhamettu Video/Image Modeling and Synthesis(VIMS) Lab, Dept. of Computer and Information Sciences Cathleen Geiger Center for Climatic Research, Department. of Geography University of Delaware, USA
Sea ice
Need for reconstruction “The feasibility of using snow surface roughness to infer ice thickness and ice bottom roughness is promising….” “…the goal of a circumpolar high resolution data set of Antarctic sea ice and snow thickness distributions has not yet been achieved …” “…crucial for future validation of satellite observations, climate models, and for assimilation into forecast models…” Ref: Workshop on Antarctic Sea Ice Thickness, 2006; Annals of Glaciology
Outline Stereo in presence of large texture-less areas Entropy based Segmentation Our Approach Two stage estimation MRF Formulation Occlusion Model Comparison of results Conclusion
Sample Images
Some characteristics in images Smoothly changing disparity No edge Low color variation
Stereo Left Image Hierarchical BP Graph Cuts Our Algorithm
Previous approaches MethodMatching measureSegmentationMultiple SegmentationHierarchicalOcclusion Model Klaus et al. ICPR 2006 Matching Pixels +Voting for plane + BP Adaptive measure with SAD and gradient Mean Shift No (Only one image) No Hong et al. CVPR 2004 Matching Pixels + Graph cutSAD, FixedMean Shift No (Only one image) NoYes Wang et al. CVPR 2008 Adaptive correlation window + Voting for plane + BP Adaptive correlationMean Shift No (Only one image) NoYes Nister et al. CVPR 2006 Matching Pixels + Hierarchical BPColor weighted correlationRectangular Grid Yes (Both images) Yes Trinh BMVC 2008 Matching Segments + BPSSD + gradientMean shift Yes (Only one image) YesNo Felzenswalb et al. CVPR 2004 Matching Pixels + BPNone-Yes Our MethodMatching Segments + 2 level BPSAD Entropy filtering + graph based segmentation No2 levelsYes
Entropy based segmentation argmax
Entropy based segmentation 1. Convert the image to grayscale and calculate the histogram. 2. Estimate the brightness threshold as the gray value that maximizes the entropy of the segmented image. 3. Partition the histogram based on that threshold into two parts. Equalize the two histograms. For each histogram repeat steps 2 and 3.
Comparison with mean shift Left Image Entropy based segments Entropy based segmentation Mean Shift segments
Our approach Two stage solution S2S2 S 3 S1S1 S2S2 S1S1 S2S2 S1S1 Segment disparity Single disparity per segment Fewer disparity levels Segment neighborhood Pixel disparity Disparity per pixel Full range of disparities Pixel neighborhood Occlusion Detection
Example
MRF Formulation Segment Level Disparity
MRF Model Pixel Level Disparity
Occlusion Model Rohith MV, Gowri Somanath, Chandra Kambhamettu, Cathleen Geiger Towards estimation of dense disparities from stereo images containing large textureless regions. 19th International Conference on Pattern Recognition(ICPR), 2008
Results
Middlebury dataset
Conclusions Entropy based segmentation to handle large texture- less regions Two step MRF formulation Solution using belief propagation Can handle large disparity ranges
Future work Explore combination of segmentations based on region characteristics Use priors over segmentation and disparity calculation in sequence of images
Acknowledgements This work was made possible by National Science Foundation (NSF) Office of Polar Program grants, ANT and ARC