Fluctuations of disparity space image for stereo matching in untextured regions

Zhu, Hongmei; Yin, Jingwei; Yuan, Di; Sui, Wei

doi:10.1109/icip.2015.7351066

Cited by 1 publication

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed CV‐based refinement strategy is actually generalisable to both cost functions and support weights. Apart from TAD C + G as in our previous work [20], we also perform the refinement strategy on other three cost functions and all the ASWs. The representative cost curves are shown in Fig.…”

Section: Resultsmentioning

confidence: 99%

“…For convenience, the pixels from large weakly textured regions are called ambiguous pixels. From our previous work [20], we can observe that cost vectors from well-textured regions usually have salient and unique extreme values, while cost vectors from large weakly textured regions vary smoothly or fluctuate periodically. Inspired by this observation, we define a criterion for ambiguity recognition which is expressed as follows:…”

Section: Cost Volume Refinementmentioning

confidence: 93%

“…For convenience, the pixels from large weakly textured regions are called ambiguous pixels. From our previous work [20], we can observe that cost vectors from well‐textured regions usually have salient and unique extreme values, while cost vectors from large weakly textured regions vary smoothly or fluctuate periodically. Inspired by this observation, we define a criterion for ambiguity recognition which is expressed as follows:

F_{a} false({C^{'}}_{p} false) = ({, \begin{matrix} right leftthickmathspace.5em \\ 1, if falsefalse| \min_{1} false({bold-italicC}^{normal′}_{p} false) - \min_{2} false({bold-italicC}^{normal′}_{p} false) falsefalse| < τ_{3} \\ 0, otherwise \end{matrix},)

where

\min_{1}

and

\min_{2}

are the first minimum and the second minimum of the cost vector

{C^{'}}_{p}

at pixel p , respectively.…”

Section: Svcv Methodsmentioning

confidence: 99%

“…When the current pixel p shares similar colour with its right and bottom neighbours,

F_{h} false(p false)

equals 1, which indicates pixel p lies within a weakly textured region. As illustrated in [20], both the spatial coherence and the colour edge can be well preserved in

F_{h}

. Thus, the parameter

P_{0}

is computed as

P_{0} = \frac{1}{N} \underset{p}{false\sum} {bold-italicF}_{h} (p),

where N is the total number of pixels in I .…”

Section: Svcv Methodsmentioning

confidence: 99%

“…Our proposed strategy belongs to this category. In our previous work [20], the effectiveness of the cost refinement strategy is only evaluated on two test datasets. In this paper, its generalisation to different cost functions and different support weights is verified through extensive experiments.…”

Section: Related Workmentioning

confidence: 99%

See 4 more Smart Citations

SVCV: segmentation volume combined with cost volume for stereo matching

Zhu

Yin

Yuan

2017

IET Computer Vision

Self Cite

View full text Add to dashboard Cite

Stereo matching between binocular stereo images is fundamental to many computer vision tasks, such as threedimensional (3D) reconstruction and robot navigation. Various structures of real 3D scenes lead stereo matching to be an old yet still challenging problem. In this study, the authors proposed a novel adaptive support weights technique which exploits the hierarchical information provided by multilevel segmentation to preserve the robustness to imaging conditions and spatial proximity in cost aggregation. Besides, a generalisable cost refinement strategy is designed to remove the matching ambiguity in large weakly textured regions. The proposed strategy utilises both the fluctuation of the filtered cost volume and the colour information to further improve the matching accuracy. Experimental results of 50 stereo images demonstrate the effectiveness and efficiency of the proposed method. Furthermore, a systematic evaluation is developed to assess the conventional steps in local stereo methods and then reliable suggestions are given to the beginners and researchers outside the stereo matching field.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Cost Volume Refinementmentioning

confidence: 93%

F_{a} false({C^{'}}_{p} false) = ({, \begin{matrix} right leftthickmathspace.5em \\ 1, if falsefalse| \min_{1} false({bold-italicC}^{normal′}_{p} false) - \min_{2} false({bold-italicC}^{normal′}_{p} false) falsefalse| < τ_{3} \\ 0, otherwise \end{matrix},)

where

\min_{1}

and

\min_{2}

are the first minimum and the second minimum of the cost vector

{C^{'}}_{p}

at pixel p , respectively.…”

Section: Svcv Methodsmentioning

confidence: 99%

“…When the current pixel p shares similar colour with its right and bottom neighbours,

F_{h} false(p false)

equals 1, which indicates pixel p lies within a weakly textured region. As illustrated in [20], both the spatial coherence and the colour edge can be well preserved in