Joint demosaicking and zooming using moderate spectral correlation and consistent edge map

A reproducible implementation.

This Matlab code package implements the Joint Demosaicking and Zooming algorithm which has been described in the following paper:
D. Zhou, W. Dong, and W. Chen; "Joint demosaicking and zooming using moderate spectral correlation and consistent edge map," J. Electron. Imaging, 23(4), 043010 (2014). doi:10.1117/1.JEI.23.4.043010.
It includes 18 functions:

bayerGR.m - Downsample an image according to the Bayer pattern
ACPIgreenH.m - Demosaick the green plane using ACPI
horzitontal direction interpolation
ACPIgreenV.m - Demosaick the green plane using ACPI
vertical direction interpolation
jointDDFWgreen.m - Demosaik the green plane and store the estimated edge
direction information in the directional matrix DM
DDFW_BG_diff - Demosaick the blue/green color difference plane
DDFW_RG_diff - Demosaick the red/green color difference plane
DDFW_refine_green - Refine the green plane
DDFW_refine_BG_diff - Refine the blue/green color difference plane
DDFW_refine_RG_diff - Refine the red/green color difference plane
DDFWweights.m - Compute the weights
jointDDFW - Demosaick a CFA mosaic image
zoomGreen - Zoom the green plane
zoomColorDiff - Zoom color difference planes
jointZoom - Zoom a color image by a factor of 2
jointDZ - Demosaick and zoom a CFA image
RoundImage.m - Round an image in the range 0~255
cpsnr_calc.m - Compute CPSNR
testJoint.m - test function for the joint demisaicking and zooming algorithm

4 folds includes:

Kodak - Original Kodak test images
McM - Original McMaster test images
K_results - Resulted images using the proposed algorithm on the Kodak set
M_results - Resulted images using the proposed algorithm on the McMaster set

To test our algorithm, you can simply run the test function: testJoint.

For further information, please contact: [e-mail]