TY - JOUR
T1 - Depth Analogy
T2 - Data-Driven Approach for Single Image Depth Estimation Using Gradient Samples
AU - Choi, Sunghwan
AU - Min, Dongbo
AU - Ham, Bumsub
AU - Kim, Youngjung
AU - Oh, Changjae
AU - Sohn, Kwanghoon
N1 - Funding Information:
This work was supported by the National Research Foundation of Korea within Ministry of Science, ICT and Future Planning through the Korean Government under Grant NRF-2013R1A2A2A01068338. The work of B. Ham was supported by the European Research Council, VideoWorld. The associate editor coordinating the review of this manuscript and approving it for publication was Dr. Karsten Mueller.
Publisher Copyright:
© 2015 IEEE.
PY - 2015/12
Y1 - 2015/12
N2 - Inferring scene depth from a single monocular image is a highly ill-posed problem in computer vision. This paper presents a new gradient-domain approach, called depth analogy, that makes use of analogy as a means for synthesizing a target depth field, when a collection of RGB-D image pairs is given as training data. Specifically, the proposed method employs a non-parametric learning process that creates an analogous depth field by sampling reliable depth gradients using visual correspondence established on training image pairs. Unlike existing data-driven approaches that directly select depth values from training data, our framework transfers depth gradients as reconstruction cues, which are then integrated by the Poisson reconstruction. The performance of most conventional approaches relies heavily on the training RGB-D data used in the process, and such a dependency severely degenerates the quality of reconstructed depth maps when the desired depth distribution of an input image is quite different from that of the training data, e.g., outdoor versus indoor scenes. Our key observation is that using depth gradients in the reconstruction is less sensitive to scene characteristics, providing better cues for depth recovery. Thus, our gradient-domain approach can support a great variety of training range datasets that involve substantial appearance and geometric variations. The experimental results demonstrate that our (depth) gradient-domain approach outperforms existing data-driven approaches directly working on depth domain, even when only uncorrelated training datasets are available.
AB - Inferring scene depth from a single monocular image is a highly ill-posed problem in computer vision. This paper presents a new gradient-domain approach, called depth analogy, that makes use of analogy as a means for synthesizing a target depth field, when a collection of RGB-D image pairs is given as training data. Specifically, the proposed method employs a non-parametric learning process that creates an analogous depth field by sampling reliable depth gradients using visual correspondence established on training image pairs. Unlike existing data-driven approaches that directly select depth values from training data, our framework transfers depth gradients as reconstruction cues, which are then integrated by the Poisson reconstruction. The performance of most conventional approaches relies heavily on the training RGB-D data used in the process, and such a dependency severely degenerates the quality of reconstructed depth maps when the desired depth distribution of an input image is quite different from that of the training data, e.g., outdoor versus indoor scenes. Our key observation is that using depth gradients in the reconstruction is less sensitive to scene characteristics, providing better cues for depth recovery. Thus, our gradient-domain approach can support a great variety of training range datasets that involve substantial appearance and geometric variations. The experimental results demonstrate that our (depth) gradient-domain approach outperforms existing data-driven approaches directly working on depth domain, even when only uncorrelated training datasets are available.
KW - 2D-to-3D conversion
KW - Depth estimation
KW - gradient transfer
KW - image analogy
KW - non-parametric sampling
UR - http://www.scopus.com/inward/record.url?scp=84960193272&partnerID=8YFLogxK
U2 - 10.1109/TIP.2015.2495261
DO - 10.1109/TIP.2015.2495261
M3 - Article
AN - SCOPUS:84960193272
SN - 1057-7149
VL - 24
SP - 5953
EP - 5966
JO - IEEE Transactions on Image Processing
JF - IEEE Transactions on Image Processing
IS - 12
M1 - 7308054
ER -