How to Deblur Color JPEG Images Using the Blind Deconvolution?

Hello,

I am going through the example of deblurring an image using blind deconvolution in the Image Processing toolbox documentation.

Throughout that example, a grayscale TIFF image is used. I would like to use a truecolor JPEG image of mine instead. Is this possible?

I was able to go through the example and create the first 6 figures, up through the 2x2 plot of the 4 different PSFs. Then I ran into problems.

The first problem I ran into was on this line:

WEIGHT = edge(I,'sobel',.3);

The error was: Error using edge Expected input number 1, I, to be two-dimensional.

Error in edge>parse_inputs (line 547)
validateattributes(I,{'numeric','logical'},{'nonsparse','2d'},mfilename,'I',1);

Error in edge (line 190)
[a,method,thresh,sigma,thinning,H,kx,ky] = parse_inputs(varargin{:});

Error in deblur_blind_deconvolution (line 55)
WEIGHT = edge(I,'sobel',.3);

I was able to get past that by changing the line that calls edge() to this:

WEIGHT = edge(rgb2gray(I),'sobel',.3);

But that gives me a different problem later on:

Error using deconvblind>parse_inputs (line 374)
If not a scalar, WEIGHT has to have size of the input image.

Error in deconvblind (line 122)
[J,P,NUMIT,DAMPAR,READOUT,WEIGHT,sizeI,classI,sizePSF,FunFcn,FunArg] = ...

Error in deblur_blind_deconvolution (line 63)
[J P] = deconvblind(Blurred,INITPSF,30,[],WEIGHT);

In this scenario, Blurred is a m x n x 3 matrix of uint8 values. WEIGHT is m x n matrix of double values.

I have found one way to get around this, but it is not satisfactory. After reading in the original image, I can convert it to grayscale right away, then just use that grayscale image throughout the deblur process (in the call to edge(), revert back to simply using I):

I = imread('mycolorjpgimage.jpg');
figure;imshow(I);title('Original Image');
I = rgb2gray(I);

But now I am stuck with a grayscale image at the end. Is there some way to convert to color and recover all of the original color data after the deblurring?

Another possibility I am thinking of is somehow converting WEIGHT from m x n to m x n x 3. But I'm stuck on how to do that.

Any help? Is deblurring a color jpeg with the WEIGHT array even possible?

(Apologies if this has been answered somewhere else already. I have googled around for hours now and have yet to find anything that addresses this specifically.)

[Walter Roberson]

You might be able to get somewhere by deblurring each color pane individually.

[KIRAN kumar]

design an individual function file using wavelets rather using predefined filters like roberts,prewitt,or sobel

[Thishan]

i am suffering from the same problem did you find an answer?

[Image Analyst]

Did you try Walter's suggestion? If not, why not?

Answer 1

[Le Chen]

Hi，I have the same problem ,do you find an answer?