Fast Cost-Volume Filtering for Visual Correspondence and Beyond

Abstract

Many computer vision tasks can be formulated as labeling problems. The desired solution is often a spatially smooth labeling where label transitions are aligned with color edges of the input image. We show that such solutions can be efficiently achieved by smoothing the label costs with a very fast edge-preserving filter. In this paper, we propose a generic and simple framework comprising three steps: 1) constructing a cost volume, 2) fast cost volume filtering, and 3) Winner-Takes-All label selection. Our main contribution is to show that with such a simple framework state-of-the-art results can be achieved for several computer vision applications. In particular, we achieve 1) disparity maps in real time whose quality exceeds those of all other fast (local) approaches on the Middlebury stereo benchmark, and 2) optical flow fields which contain very fine structures as well as large displacements. To demonstrate robustness, the few parameters of our framework are set to nearly identical values for both applications. Also, competitive results for interactive image segmentation are presented. With this work, we hope to inspire other researchers to leverage this framework to other application areas.

Direct Digital Surface Model Generation by Semi-Global Vertical Line Locus Matching

“…Binocular stereo matching is intrinsically constrained with geometric projection due to the usage of rectified images which simplifies the matching task to a two frame correspondence and inspires a large number of algorithms [1][2][3][4][5][6][7][8][9][10]. Generally, binocular stereo matching consists of four steps [1]: cost calculation, cost aggregation, disparity optimization, and refinement.…”

“…However, most of the researches focus on the cost aggregation and disparity optimization in recent years. According to the disparity optimization, local or non-local methods [2][3][4][5]8,9], which employ the cost aggregation followed by the "Winner Takes All" principle, and global methods [6,10], which tend to use the pixel-wise or object-wise cost function optimized by minimization of an energy function in the Markov Random Field (MRF) have been developed. However, most of state-of-the-art methods suffer common difficulties in certain…”

Superpixel-based graph cuts for accurate stereo matching

“…As for cost-volume filtering, given the raw cost volume ( ) computed for a certain disparity at pixel , then the filtered cost value can be computed as ̃( ) = ∑ ( ) . Here, different edge-aware filtering technologies [14,11,8] can be applied and the difference between them lies primarily in defining . Yoon [20] performs an adaptive support window technique for cost filtering, which solves boundary issues successfully, but it costs O(|W|) computation.…”

Revisiting Intrinsic Curves for Efficient Dense Stereo Matching

“…In order to preserve depth discontinuities, edge-aware weighted filters, e.g. geodesic distance, bilateral and guided filtering, can be used at the cost of noticeably higher computational complexity (Yoon and Kweon, 2006;Hosni et al, 2009;Hosni et al, 2013). The main drawback of local techniques is that they require evaluating the full disparity space image (DSI).…”