Limits...
Computationally efficient locally adaptive demosaicing of color filter array images using the dual-tree complex wavelet packet transform.

Aelterman J, Goossens B, De Vylder J, Pi┼żurica A, Philips W - PLoS ONE (2013)

Bottom Line: By using a novel locally adaptive approach for demosaicing (complex) wavelet coefficients, we show that many of the common demosaicing artifacts can be avoided in an efficient way.Results demonstrate that the proposed method is competitive with respect to the current state of the art, but incurs a lower computational cost.The wavelet approach also allows for computationally effective denoising or deblurring approaches.

View Article: PubMed Central - PubMed

Affiliation: IPI-TELIN-IMINDS, Ghent University, Ghent, Belgium. jaelterm@telin.ugent.be

ABSTRACT
Most digital cameras use an array of alternating color filters to capture the varied colors in a scene with a single sensor chip. Reconstruction of a full color image from such a color mosaic is what constitutes demosaicing. In this paper, a technique is proposed that performs this demosaicing in a way that incurs a very low computational cost. This is done through a (dual-tree complex) wavelet interpretation of the demosaicing problem. By using a novel locally adaptive approach for demosaicing (complex) wavelet coefficients, we show that many of the common demosaicing artifacts can be avoided in an efficient way. Results demonstrate that the proposed method is competitive with respect to the current state of the art, but incurs a lower computational cost. The wavelet approach also allows for computationally effective denoising or deblurring approaches.

Show MeSH
Color corruption can be caused either by excess luminance bandwidth in the vertical or horizontal direction.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3643977&req=5

pone-0061846-g007: Color corruption can be caused either by excess luminance bandwidth in the vertical or horizontal direction.

Mentions: Other artifacts relate to the low bandwidth reconstruction of chroma, of the total bandwidth. As these artifacts are psycho-visually not disturbing, which is strongly related to the efficacy of chroma subsampling in compression schemes [25], we will not take special measures to correct them. Eliminating the aforementioned artifacts consists of two steps: detecting demosaicing artifacts (Section 2.2) and correcting them (Section 2.2 and Section 2.2). The strategy to correct these artifacts is based on the redundant information in the demosaicing equations (14), which is in turn related to the existence of multiple aliasing copies of the chrominance signals, as they are subsampled both vertically and horizontally in the mosaicing process. If two uncorrupted aliasing copies of the chrominance signals can be found in the spectral content of any of the mosaic (complex) wavelet coefficients, artifact-free reconstruction is possible. Consider an image patch with some vertical stripes, i.e. large horizontal bandwidth (e.g. the picture in Figure 6). Now the luminance bandwidth assumption (12) is no longer correct. For our two level wavelet packet transformation we can express this as:(23)As a result, the simplified equations from which the demosaicing rules are derived (13) are also incorrect, and instead become, for the first of the four trees of our two level 2D wavelet packet transformation:(24)The wavelet coefficient , which should only contain chrominance alias, now contains excess luminance energy and is considered corrupted. In a demosaicing algorithm that is oblivious to these artifacts, such as when applying the algorithm in Table 2 separately to the different complex wavelet trees, the result would show a discoloration artifact. The red, as well as the blue, low-pass signal is corrupted:It is equally possible that the artifact was caused by horizontal stripes, i.e. large vertical bandwidth. Before an attempt can be made to suppress the artifact, it should be known whether the artifact is caused by excess luminance bandwidth in the horizontal direction (, graphically in Figure 7, left) or by excess luminance bandwidth in the vertical direction (, graphically in Figure 7, right). In theory, it could happen that the diagonal wavelet coefficient is corrupted (), but since this represents a higher bandwidth than the vertical and the horizontal wavelet coefficient, we do not consider this here.


Computationally efficient locally adaptive demosaicing of color filter array images using the dual-tree complex wavelet packet transform.

Aelterman J, Goossens B, De Vylder J, Pi┼żurica A, Philips W - PLoS ONE (2013)

Color corruption can be caused either by excess luminance bandwidth in the vertical or horizontal direction.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3643977&req=5

pone-0061846-g007: Color corruption can be caused either by excess luminance bandwidth in the vertical or horizontal direction.
Mentions: Other artifacts relate to the low bandwidth reconstruction of chroma, of the total bandwidth. As these artifacts are psycho-visually not disturbing, which is strongly related to the efficacy of chroma subsampling in compression schemes [25], we will not take special measures to correct them. Eliminating the aforementioned artifacts consists of two steps: detecting demosaicing artifacts (Section 2.2) and correcting them (Section 2.2 and Section 2.2). The strategy to correct these artifacts is based on the redundant information in the demosaicing equations (14), which is in turn related to the existence of multiple aliasing copies of the chrominance signals, as they are subsampled both vertically and horizontally in the mosaicing process. If two uncorrupted aliasing copies of the chrominance signals can be found in the spectral content of any of the mosaic (complex) wavelet coefficients, artifact-free reconstruction is possible. Consider an image patch with some vertical stripes, i.e. large horizontal bandwidth (e.g. the picture in Figure 6). Now the luminance bandwidth assumption (12) is no longer correct. For our two level wavelet packet transformation we can express this as:(23)As a result, the simplified equations from which the demosaicing rules are derived (13) are also incorrect, and instead become, for the first of the four trees of our two level 2D wavelet packet transformation:(24)The wavelet coefficient , which should only contain chrominance alias, now contains excess luminance energy and is considered corrupted. In a demosaicing algorithm that is oblivious to these artifacts, such as when applying the algorithm in Table 2 separately to the different complex wavelet trees, the result would show a discoloration artifact. The red, as well as the blue, low-pass signal is corrupted:It is equally possible that the artifact was caused by horizontal stripes, i.e. large vertical bandwidth. Before an attempt can be made to suppress the artifact, it should be known whether the artifact is caused by excess luminance bandwidth in the horizontal direction (, graphically in Figure 7, left) or by excess luminance bandwidth in the vertical direction (, graphically in Figure 7, right). In theory, it could happen that the diagonal wavelet coefficient is corrupted (), but since this represents a higher bandwidth than the vertical and the horizontal wavelet coefficient, we do not consider this here.

Bottom Line: By using a novel locally adaptive approach for demosaicing (complex) wavelet coefficients, we show that many of the common demosaicing artifacts can be avoided in an efficient way.Results demonstrate that the proposed method is competitive with respect to the current state of the art, but incurs a lower computational cost.The wavelet approach also allows for computationally effective denoising or deblurring approaches.

View Article: PubMed Central - PubMed

Affiliation: IPI-TELIN-IMINDS, Ghent University, Ghent, Belgium. jaelterm@telin.ugent.be

ABSTRACT
Most digital cameras use an array of alternating color filters to capture the varied colors in a scene with a single sensor chip. Reconstruction of a full color image from such a color mosaic is what constitutes demosaicing. In this paper, a technique is proposed that performs this demosaicing in a way that incurs a very low computational cost. This is done through a (dual-tree complex) wavelet interpretation of the demosaicing problem. By using a novel locally adaptive approach for demosaicing (complex) wavelet coefficients, we show that many of the common demosaicing artifacts can be avoided in an efficient way. Results demonstrate that the proposed method is competitive with respect to the current state of the art, but incurs a lower computational cost. The wavelet approach also allows for computationally effective denoising or deblurring approaches.

Show MeSH