Constant Time Weighted Median Filtering for Stereo Matching and Beyond

By Kaiming He, Yichen Wei, Enhua Wu. IEEE International Conference on Computer Vision, 2013

303 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/4a7b14b2847e140a2885b41797164a812adcd4ad

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Rolling Guidance Filter (2014, background): “They include representatives of anisotropic diffusion [20], bilateral filter (BF) [26], guided filter (GF) [13], geodesic filters [7,11], weighted median filters [18,34], to name a few.”
D-HAZY: A dataset to evaluate quantitatively dehazing algorithms (2016, unknown): “[31].”
The leaf angle distribution of natural plant populations: assessing the canopy with a novel software tool (2015, background): “The other three optional filters correct for (i) outliers with a weighted median filter [50], (ii) missing pixels via filling of small gaps, (iii) undersized disparity fragments by removing them.”
The Fast Bilateral Solver (2015, methodology): “Examples of this include joint bilateral ﬁltering [37,40] and upsampling [20], adaptive manifolds [12], the domain transform [11], the guided ﬁlter [17,16], MST-based ﬁltering [41], and weighted median ﬁltering [30,44].”
Robust depth estimation for light field via spinning parallelogram operator (2016, unknown): “The initial disparity after guided filter is then optimized using weighted median filter [30], graph cuts [31], and finally the iterative refinement [32].”
Accurate depth map estimation from a lenslet light field camera (2015, methodology): “As a sequential step, every cost slice is reﬁned using an edge-preserving ﬁlter [15] to alleviate the coarsely scattered unreliable matches.”
A primal-dual framework for real-time dense RGB-D scene flow (2015, methodology): “As an alternative, we opt for using a 3 × 3 weighted median filter, similar to the one described in [23].”
3D Photography Using Context-Aware Layered Depth Inpainting (2020, methodology): “We, therefore, sharpen the depth maps using a bilateral median filter [37] (Figure 2d), using a 7×7 window size, and σspatial = 4.”
Soft 3D reconstruction for view synthesis (2017, methodology): “2003] seek to minimize a global cost function that also optimizes for depth smoothness, while ‘local’ methods attempt to achieve the same more efficiently (or even at real-time rates) using patch-based O(1) edge-aware filtering kernels such as [Hosni et al. 2011; Lu et al. 2013; Ma et al. 2013].”
Fast Image Processing with Fully-Convolutional Networks (2017, background): “Another family of dedicated acceleration schemes addresses the median filter and its variants [72, 61, 54, 80].”
Multi-View Stereo: A Tutorial (2015, background): “Although bilateral filters are the most common, other anisotropic filters can be used for stereo such as a weighted median filter [137].”
Optical flow modeling and computation: A survey (2015, methodology): “The combination of the three ingredients has led to the development of competitive optical ﬂow estimation methods based on pure feature matching locally ﬁltered [122,172,238].”
Displets: Resolving stereo ambiguities using object knowledge (2015, background): “Local stereo methods [14, 20, 23, 27, 37, 46, 50, 67] often fail in challenging scenarios as they suffer from matching ambiguities in weakly textured, saturated or reflective regions.”