Stereo Matching with Improved Radiometric Invariant Matching Cost and Disparity Refinement

Shi, Jinjin; Fu, Fang; Wang, Yao; Xu, Weizhe; Wang, Jinxiang

doi:10.1007/978-3-319-42291-6_7

Cited by 3 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Each image was converted into a lab color space, and l channel was utilized as input data. The proposed QCT was compared with the popular census transform [10], AD-Census [12], RG-Census [13], TCC-Census [14], and 4M-Census [17] proposed recently.…”

Section: Resultsmentioning

confidence: 99%

“…However, only a few studies have been conducted to find proper patch sizes for these methods. The census transform [10], AD-Census [12], and RG-Census [13] simply apply a square patch with a fixed size as shown in Fig. 5(a).…”

Section: B Variable-sized Patch (Vsp)mentioning

confidence: 99%

“…Mei et al combined the census transform with the conventional pixel-wise absolute difference (AD-Census) to take the advantages of both [12]. Shi et al applied the census transform in the relative gradient [7] domain instead of the conventional spatial domain (RG-Census) [13]. Chang et al proposed a modified version of the census transform called a trinary cross color census transform (TCC-Census) [14].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Quaternary Census Transform Based on the Human Visual System for Stereo Matching

Kim

Lim

et al. 2020

IEEE Access

View full text Add to dashboard Cite

The census transform is a non-parametric local transform that is widely used in stereo matching. This transform encodes the structural information of a local patch into a binary code stream representing the relative intensity ordering of the pixels within the patch. Despite its high performance in stereo matching, the census transform often generates identical binary code streams for two different patches because it simply thresholds the pixels within the patch at the center pixel intensity. To overcome this problem, we introduce a quaternary census transform that encodes the local structural information into a quaternary code stream by employing both the relative intensity ordering and the minimum visibility threshold of the human eye known as the just-noticeable difference. Moreover, because the human eye activates different areas of the retina based on brightness, the patch size for the proposed quaternary census transform adaptively varies depending on the luminance of each pixel. Experimental results on well-known Middlebury stereo datasets prove that the proposed transform outperforms the other census transform-based methods in terms of the accuracy of stereo matching. INDEX TERMS Census transform, depth estimation, disparity map, human visual system, similarity cost calculation, stereo image processing, stereo matching.

show abstract