Depth estimation from image structure

IEEE Transactions on Pattern Analysis and Machine Intelligence - Tập 24 Số 9 - Trang 1226-1238 - 2002
A. Torralba1, A. Oliva2
1Artifical Intelligence Laboratory, NE 43-743 MIT, Cambridge, MA, USA
2Center for Ophthalmic Research, Brigham and Women's Hospital, Boston, MA, USA

Tóm tắt

In the absence of cues for absolute depth measurements as binocular disparity, motion, or defocus, the absolute distance between the observer and a scene cannot be measured. The interpretation of shading, edges, and junctions may provide a 3D model of the scene but it will not provide information about the actual "scale" of the space. One possible source of information for absolute depth estimation is the image size of known objects. However, object recognition, under unconstrained conditions, remains difficult and unreliable for current computational approaches. We propose a source of information for absolute depth estimation based on the whole scene structure that does not rely on specific objects. We demonstrate that, by recognizing the properties of the structures present in the image, we can infer the scale of the scene and, therefore, its absolute mean depth. We illustrate the interest in computing the mean depth of the scene with application to scene recognition and object detection.

Từ khóa

#Layout #Motion measurement #Information resources #Object recognition #Image recognition #Object detection

Tài liệu tham khảo

10.1016/S0042-6989(97)00008-4 10.1023/A:1026553619983 rogowitz, 1998, Perceptual Image Similarity Experiments, Proc SPIE Conf Human Vision and Electronic Imaging palmer, 1999, Vision Science 10.1038/381607a0 10.1109/TPAMI.1984.4767591 papoulis, 1984, Probability random variables and stochastic processes 10.1109/34.506794 10.1023/A:1011139631724 oliva, 1999, Global Semantic Classification of Scenes Using Power Spectrum Templates, Proc Challenge of Image Retrieval 10.1023/A:1007925832420 10.1016/S0364-0213(99)80027-5 10.1016/0004-3702(81)90021-7 lindeberg, 1998, Principles for Automatic Scale Selection, Int'l J Computer Vision, 30, 77 gershnfeld, 1999, The Nature of Mathematical Modeling 10.1007/BF01469346 10.1364/JOSAA.4.002379 lee, 2001, Occlusion Models for Natural Images: A Statistical Study of a Scale-Invariant Dead Leaves Model, Int'l J Computer Vision, 41, 35, 10.1023/A:1011109015675 de bonet, 1997, Structure Driven Image Database Retrieval, Advances in Neural Information Processing, 10, 866 10.1109/TPAMI.1987.4767956 10.1109/ICCV.1999.790349 10.1109/IVL.1997.629719 10.1016/S0031-3203(98)00079-X bergen, 1991, Computational Modeling of Visual Texture Segregation, Computational Models of Visual Processing, 253 10.1109/ICCV.2001.937604 10.1109/ICCV.2001.937654 10.1109/ICCV.1999.790424 10.1088/0954-898X_3_1_008 10.1109/CAIVD.1998.646032 10.1109/ICPR.1994.576325 10.1109/34.385983 10.1146/annurev.neuro.24.1.1193 10.1109/ICIP.1995.537667 10.1162/neco.1994.6.2.181 shimshoni, 2000, Shape Reconstruction of 3D Bilaterally Symmetric Surfaces, Int'l J Computer Vision, 2, 1 10.1017/CBO9780511984037.004 10.1023/A:1008120406972 horn, 1989, Shape from Shading 10.1016/0042-6989(96)00002-8 10.1145/218380.218446