Adaptive Demosaicking

Digital Still Color Cameras sample the visible spectrum using an array of color filters overlaid on a CCD such that each pixel samples only one color band. The resulting mosaic of color samples is processed to produce a high resolution color image such that a value of a color band not sampled at a certain location is estimated from its neighbors. This is often referred to as "demosaicking". The human retina has a similar structure although the distribution of cones is not as regular. Motivated by the human visual system, we propose an adaptive demosaicking technique in the framework of bilateral filtering. This approach provides us with a means to denoise, sharpen and demosaic the image simultaneously. The proposed method along with a variety of existing demosaicking strategies are run on synthetic images and real-world images for comparative purposes. A recently proposed image comparison measure geared specifically towards demosaicking has also been applied on these images to provide a performance measure.

The primary citation for this work is as follows:

R. Ramanath and Wesley E. Snyder, "Adaptive Demosaicking," to appear in the Journal of Electronic Imaging, vol. 12, no. 4, 2003.

@ARTICLE(Ramanath:03,
author = "R. Ramanath and W. E. Snyder",
title = "Adaptive Demosaicking",
journal = "to appear in Journal of Electronic Imaging",
year = "2003",
volume = "12",
number = "4",
pages = ""
)

 

The figures below are the color figures from the above manuscript. They have been provided only for web-reference purposes and have been saved in JPEG format with an sRGB color profile. In case you want a non-compressed format, please contact the person listed at the end of this page. This page is best-viewed in a screen with greater than 1024x768 resoultion -- this will permit the reader to view all the sub-figures in the same screen without having to scroll.

 

(a)

(b)
FIG. 1. Comparison of (a) Bayer Color Filter Array and (b) the distribution of cones in the human retina

 

 
FIG. 4. Bilateral filtering using the \Delta E*ab metric (a) Original Image. (b) image corrupted by Gaussian Noise (c) blur kernel (d) similarity kernel at row=50, col=51 (e) combined kernel (f) resulting image

 

 
FIG. 5. Bilateral .ltering using the .E*ab metric (a) Original Image. (b) image corrupted by Gaussian Noise (c) blur kernel (d) similarity kernel at row=50, col=51 (e) combined kernel (f) resulting image

 

 


(a)

(b)
FIG. 6. Bayer Array (a)Red channel (b)Green channel

 

 


(a)

(b)

(c)

(d)

(e)
 
FIG. 11. Original Images used in this experiment (a) Test Image1 (b) Crayon image (c) Hibiscus image (d) Region of interest (ROI) in the Crayon image (e) ROI in the Hibiscus image

 

 


(a)

(b)

(c)

(d)

(e)

(f)
FIG. 12. Results using on Test Image_1, with SNR 30dB (a) Linear (b) Cok (c) Freeman (d) Bilateral_1 with \sigma_s = 0.5 (e) Bilateral_2 with \sigma_s = 0.5 (f) Bilateral_3 with \sigma_s = 20

 

 


(a)

(b)

(c)

(d)

(e)

(f)
FIG. 13. Results using on Crayon image, with SNR 30dB (a) Linear (b) Cok (c) Freeman (d) Bilateral_1 with \sigma_s = 0.5 (e) Bilateral_2 with \sigma_s = 0.5 (f) Bilateral_3 with \sigma_s = 20

 

 


(a)

(b)

(c)

(d)

(e)

(f)
FIG. 13. Results using on Hibiscus image, with SNR 30dB (a) Linear (b) Cok (c) Freeman (d) Bilateral_1 with \sigma_s = 0.5 (e) Bilateral_2 with \sigma_s = 0.5 (f) Bilateral_3 with \sigma_s = 20

 

Contact rramana@eos.ncsu.edu for more information regarding methods and datasets.

The Crayon image has been provided by Phil Askey. His donation is graciously acknowledged.