360 likes | 471 Vues
This paper presents an efficient near real-time stereo matching technique using geodesic diffusion for enhanced depth perception. The method combines anisotropic diffusion and geodesic diffusion to improve matching accuracy and speed. Experimental results show promising outcomes with CUDA implementation for faster processing.
E N D
Near Real-Time Stereo Matching Using Geodesic Diffusion Leonardo De-Maeztu, Arantxa Villanueva, Member, IEEE, and Rafael Cabeza IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 34, NO. 2, FEBRUARY 2012 • Guan-Yu Liu
Outline • Introduction • Overview • Related work • Method • Experimental Results • CUDA • Q & A
Introduction(1/4) • Stereo matching • Local matching • A finite region(window size) is being computed • Global matching • Do smoothness by energy minimization techniques
Introduction(2/4) • When using local support regions, it is implicitly assumed that all pixels in the region are of the same depth. • the fronto-parallel surfaces assumption • Adaptive-weight methods
Introduction(3/4) • Adaptive-weight methods are the local algorithms yielding the best results. • Highly time-consuming task • Anisotropic diffusion, a computer vision technique very similar to adaptive weighting but computationally less expensive. • a computer vision technique very similar to adaptive weighting but computationally less expensive.
Introduction(4/4) • Geodesic diffusion is inspired by anisotropic diffusion. • diffusing both matching costs and weights. • Near real-time execution is demonstrated using a commercial graphics card.
Related Work • Adaptive-weight methods [7] • Adaptive-weight methods [8] • Anisotropic diffusion [9] [7] K.-J. Yoon and I.S. Kweon, “Adaptive Support-Weight Approach for Correspondence Search,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 28, no. 4, pp. 650-656, Apr. 2006. [8] A. Hosni, M. Bleyer, M. Gelautz, and C. Rhemann, “Local Stereo Matching Using Geodesic Support Weights,” Proc. Int’l Conf. Image Processing, pp. 2093-2096, 2009. [9] P. Perona and J. Malik, “Scale-Space and Edge Detection Using Anisotropic Diffusion,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 12, no. 7, pp. 629-639, July 1990.
Related Work(1/6) • Adaptive-weight methods [7]
Related Work(2/6) • Adaptive-weight methods [7] • 123 • 123 • 123 Euclidean distance between the values in the CIELab color space and spatial euclidean distance truncated absolute difference (TAD)
Related Work(3/6) • Adaptive-weight methods [8]
Related Work(4/6) • Adaptive-weight methods [8] • 123 • 123 • 123 Shortest path
Related Work(5/6) • The two algorithms use the same optimization technique, winner-takes-all (WTA).
Related Work(6/6) • Anisotropic diffusion is a computer vision technique similar to bilateral filtering. • only the comparison of each pixel with its immediate neighbors is necessary.
Method • A : Anisotropic diffusion • B : Geodesic diffusion
Method.A(1/3) • Anisotropic diffusion
Method.A(2/3) • Anisotropic diffusion • 123 • 123 Euclidean distance between the values in the CIELab color space
Method.A(3/3) • It is aniterative computer vision technique.[9]
Method • A : Anisotropic diffusion • B : Geodesic diffusion
Method.B(1/8) • Three principles • Costs and weights are diffused so that the importance of each cost value is known in each iteration. • In each iteration, the costs and weights at each pixel are accumulated. After the last iteration, all the support region information has been accumulated at each pixel. • To increase the efficiency of information diffusion and to avoid loops, turns in the direction of diffusion are penalized.
Method.B(2/8) • Geodesic diffusion
Method.B(3/8) • Each of the four positions inherits the costs and weights of each of the four direct neighbors of each pixel.
Method.B(4/8) • Geodesic diffusion • 123 • 123 • 123
Method.B(5/8) • i= 0 right neighbors • i = 1 lower neighbors • i = 2 upper neighbors • i = 3 left neighbors
Method.B(6/8) • The cost and weight information derived from a direct neighbor is not returned to this neighbor. • Costs are only propagated with their full weights in the same direction of their propagation direction in the previous iteration.
Method.B(7/8) • Geodesic diffusion • 123 • 123
Method.B(8/8) • At the end of the diffusion process, the DSI costs are normalized. • Thus, concluded, and the disparity map is then computed by selecting the lower cost disparity for each pixel WTA.
CUDA • CUDA implementation of our algorithm ran in less than 60 milliseconds for the Tsukuba stereo pair on a GeForce 480 GTX card.