Gradient-guided filtering of depth maps using deep neural networks

Ochotorena, Cecille Adrianne; Ochotorena, Carlo Noel; Dadios, Elmer P.

doi:10.1109/hnicem.2015.7393265

Cited by 5 publications

(4 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As shown in Fig. 4, especially in the red box, we can see that edges extracted from the depth map interpolated by bicubic contains obvious jagged edges, while the edge maps of depth images after using guided filter and gradient domain guided filter [38], [39] cannot be recognized. It shows our result is most similar to the ground graph relatively.…”

Section: B Color Guidance With High-resolution Color Imagementioning

confidence: 96%

“…Although GF [18] could keep the edges well and compute easily, it also suffered from halo artifacts sometimes. Gradient domain guided filter [38], [39] could keep edge better by adding an explicit first order edge sensing constraint. Joint Bilateral Filter (JBF) [29] employed an additional guidance to improve the quality of the input target image taken from a dark or noisy environment.…”

Section: B Depth Image Super-resolutionmentioning

confidence: 99%

See 1 more Smart Citation

Deep Color Guided Coarse-to-Fine Convolutional Network Cascade for Depth Image Super-Resolution

Wen

Sheng

et al. 2019

IEEE Trans. on Image Process.

105

View full text Add to dashboard Cite

Depth image super-resolution is a significant yet challenging task. In this paper, we introduce a novel deep color guided coarse-to-fine convolutional neural network (CNN) framework to address this problem. First, we present a datadriven filter method to approximate the ideal filter for depth image super-resolution instead of hand-designed filters. Based on large data samples, the filter learned is more accurate and stable for upsampling depth image. Second, we introduce a coarse-to-fine CNN to learn different sizes of filter kernels. In coarse stage, larger filter kernels are learned by CNN to achieve crude high-resolution depth image. As to fine stage, the crude high-resolution depth image is used as the input so that smaller filter kernels are learned to gain more accurate results. Benefit from this network, we can progressively recover the high frequency details. Third, we construct a color guidance strategy that fuses color difference and spatial distance for depth image upsampling. We revise the interpolated high-resolution depth image according to the corresponding pixels in highresolution color maps. Guided by color information, the depth of high-resolution image obtained can alleviate texture copying artifacts and preserve edge details effectively. Quantitative and qualitative experimental results demonstrate our state-of-the-art performance for depth map super-resolution.

show abstract

Section: B Color Guidance With High-resolution Color Imagementioning

confidence: 96%

Section: B Depth Image Super-resolutionmentioning

confidence: 99%

Deep Color Guided Coarse-to-Fine Convolutional Network Cascade for Depth Image Super-Resolution

Wen

Sheng

et al. 2019

IEEE Trans. on Image Process.

105

View full text Add to dashboard Cite

show abstract

“…Adaptively some researchers are utilising learning‐based approaches for depth map enhancement. In [67], a colour‐guided multi‐layer NN is designed based on staked denoising auto‐encoders to enhance distorted depth maps. Zhang et al [69] also propose a deep convolutional NN (CNN) with a pre‐processing step.…”

Section: Depth Map Artefacts Reductionmentioning

confidence: 99%

Depth map artefacts reduction: a review

et al. 2020

View full text Add to dashboard Cite

“…Therefore, some researchers have used learning methods to filter and enhance depth maps. For example, Ochotorena et al [47] proposed a method based on neural networks (NNs) whereby they built a colour‐guided multi‐layer NN using stacked denoising autoencoders for depth map filtering. Zhang and Wu [48] proposed a deep convolutional NN (CNN)‐based method for depth map enhancement with a pre‐processing step.…”

Section: Related Work and Motivationmentioning

confidence: 99%