Richardson-Lucy 2D deconvolution

image =	richardsonLucyDeconvolution2dImg (inRealImg,inKnlXY,nbIter)
image =	richardsonLucyDeconvolution2dImg (inRealImg,inKnlXY,nbIter,bStaturateImg)
image =	richardsonLucyDeconvolution2dImg (inRealImg,inKnlXY,nbIter,epsilonValue)
image =	richardsonLucyDeconvolution2dImg (inRealImg,inKnlXY,nbIter,epsilonValue,bStaturateImg)

Detailed Description

Richardson-Lucy deconvolution algorithm.

This algorithm removes iteratively blur from an image according to a given point spread function. The implementation is inspired from https://scikit-image.org/docs/dev/api/skimage.restoration.html#skimage.restoration.richardson_lucy.

Note

The input image must be normalized.

For a given iteration , the deconvolution result is given by :

Where is the deconvolved image at iteration , is the input PSF, is the flipped (mirrored along every directions) PSF and is the convolution operator.

Contrary to Richardson-Lucy deblur 2d, the input PSF is an IPSDK kernel, which offers more flexibility. It is also possible to define an epsilon value to determine from which value is considered as 0. In that case, is set to 0. Finally, by saturating the output image, all values below -1 and higher than 1 are saturated in the range [-1, 1].

Here is a result of Richardson-Lucy deconvolution with 20 iterations a 15x15 mean kernel:

See also

https://en.wikipedia.org/wiki/Richardson%E2%80%93Lucy_deconvolution

Example of Python code :

Example imports

import PyIPSDK
import PyIPSDK.IPSDKIPLFiltering as filter

Code Example

    # opening of input images
    geom = PyIPSDK.geometry2d(PyIPSDK.eImageBufferType.eIBT_Real32, 510, 509)
    inImg = PyIPSDK.loadRawImageFile(inputImgPath, geom)
    
    # create a processing kernel
    inKnl = PyIPSDK.rectangularKernelXY(1, 1, [1/9, 1/9, 1/9, 1/9, 1/9, 1/9, 1/9, 1/9, 1/9])
    
    # deconvolution filter 2d computation without normalization
    outImg = filter.richardsonLucyDeconvolution2dImg(inImg, inKnl, 50)

    # deconvolution filter 2d computation with normalization
    outNormalizedImg = filter.richardsonLucyDeconvolution2dImg(inImg, inKnl, 50, True)

Example of C++ code :

Example informations

Header file

#include <IPSDKIPL/IPSDKIPLFiltering/Processor/RichardsonLucyDeconvolution2dImg/RichardsonLucyDeconvolution2dImg.h>

Code Example

        // opening input image
        ImagePtr pInImg = loadRawImageFile(inputImgPath, *pGeometry);
        
        // compute un-normalized deconvolution on input image
        ImagePtr pOutImg = richardsonLucyDeconvolution2dImg(pInImg, pInKnlXY, 50);
        ImagePtr pOutImg_normalized = richardsonLucyDeconvolution2dImg(pInImg, pInKnlXY, 50, true);