I need help on cell array and Imaging processing

Hi

I am using the following code

close all;
clc; 
 
%%  Initalize the data 
 
dataDir= fullfile('Data/');
exts = {'.jpg','.png','.tif','BMP'};
 
 
imds = imageDatastore(fullfile(dataDir),...
'IncludeSubfolders',true,'FileExtensions','.jpg','LabelSource','foldernames');
countEachLabel(imds);
 
[TrainData, TestData] = splitEachLabel(imds, 0.5);
size(TrainData);
countEachLabel(TrainData);
numImages = numel(TrainData.Files); 
for i = 1:numImages
    img = readimage(TrainData, i);
%     img2= imshow(img, 'InitialMagnification', 800);
    img3= imresize(img, [100 100]);
    img4= imshow(img3, 'InitialMagnification', 800);
     drawnow;
    Train{i} = (img3);         %#ok<SAGROW>
 
end
hiddenSize = 25;
autoenc = trainAutoencoder(Train,hiddenSize,'MaxEpochs',1000,...
'DecoderTransferFunction','purelin','EncoderTransferFunction','satlin','L2WeightRegularization',0.0004,'SparsityRegularization',4,'SparsityProportion',0.15);
 
numImages = numel(TestData.Files); 
for i = 1:numImages
    img5 = readimage(TestData, i);
    img6= imresize(img, [100 100]);
    img7= imshow(img3, 'InitialMagnification', 800);
     drawnow;
    Test{i} = (img6);      %#ok<SAGROW>
end
xReconstructed = predict(autoenc,Test);
%% Test Images 
figure();
for i = 1:10
    subplot(4,5,i);
    imshow(TestData.Files{i});
end
 
%% Reconstructed images from TestData
figure();
for i = 1:10
    subplot(4,5,i);
    imshow(xReconstructed(i))
end

But I got the following error

Error using imageDisplayValidateParams
Expected input number 1, I, to be one of these types:
 
double, single, uint8, uint16, uint32, uint64, int8, int16, int32, int64, logical
 
Instead its type was cell.
 
Error in images.internal.imageDisplayValidateParams (line 11)
validateattributes(common_args.CData, {'numeric','logical'},...
 
Error in images.internal.imageDisplayParseInputs (line 79)
common_args = images.internal.imageDisplayValidateParams(common_args);
 
Error in imshow (line 253)
    images.internal.imageDisplayParseInputs({'Parent','Border','Reduce'},preparsed_varargin{:});
 
Error in data_process1 (line 51)
    imshow(xReconstructed(i))

When I compare to the example,

https://www.mathworks.com/help/deeplearning/ref/trainautoencoder.html

the issue is with Train{i}and Test {i}, The model is not trained probably with given data. When I compared with example in MATLAB for digitTrainCellArrayData and digitTTestCellArrayData I am having problems and my cell array which are not the same as in example

My cell arrays for Train{i}and Test {i} are like this like for example Train{i}

But in the example in MATLAB for digits, digitTrainCellArrayData and digitTestCellArrayData are like this

digitTrainCellArrayData

In addition to that. Range in example is from 0 – 1 while my data is from 0 – 255

Kindly looking for your support

Answer 1

[drummer]

figure();
for i = 1:10
    subplot(4,5,i);
    imshow(TestData.Files{i});
end

Hi.

It's probably your argument for imshow that is cell type, while it only accepts the type mentioned in your error message.

Try this workaround

figure();
for i = 1:10
    subplot(4,5,i);
    image = cell2mat(TestData.Files{i});
    imshow(image);
end

I haven't tested, so before performing within the loop, try outside, with a single image.

Cheers

[Abdussalam Elhanashi]

Hi drummer

Thanks for yur reply

I have problems with commands

Train{i} = (img3);  
Test{i} = (img6); 

I am not converting the images into correct matrix required model

I should have 1-by-102 cell array, where each cell containing a 28-by-28 matrix representing a synthetic image. like this with values 0 to 1

but i having this with wrong range 0 255

[Walter Roberson]

im2double() to convert [0 255] range to [0 1] range

[Abdussalam Elhanashi]

Hi Walter

I am using the following code

close all;
clc; 
 
%%  Initalize the data 
 
dataDir= fullfile('Data/');
exts = {'.jpg','.png','.tif','BMP'};
 
 
imds = imageDatastore(fullfile(dataDir),...
'IncludeSubfolders',true,'FileExtensions','.jpg','LabelSource','foldernames');
countEachLabel(imds);
 
[TrainData, TestData] = splitEachLabel(imds, 0.5);
size(TrainData);
countEachLabel(TrainData);
numImages = numel(TrainData.Files); 
for i = 1:numImages
    img = readimage(TrainData, i);
%     img2= imshow(img, 'InitialMagnification', 800);
    img3= imresize(img, [100 100]);
    img4= imshow(img3, 'InitialMagnification', 800);
     drawnow;
    Train{i} = (img3);         %#ok<SAGROW>
 
end
hiddenSize = 25;
autoenc = trainAutoencoder(Train,hiddenSize,'MaxEpochs',1000,...
'DecoderTransferFunction','purelin','EncoderTransferFunction','satlin','L2WeightRegularization',0.0004,'SparsityRegularization',4,'SparsityProportion',0.15);
 
numImages = numel(TestData.Files); 
for i = 1:numImages
    img5 = readimage(TestData, i);
    img6= imresize(img, [100 100]);
    img7= imshow(img3, 'InitialMagnification', 800);
     drawnow;
    Test{i} = (img6);      %#ok<SAGROW>
end
xReconstructed = predict(autoenc,Test);
%% Test Images 
figure();
for i = 1:10
    subplot(4,5,i);
    imshow(TestData.Files{i});
end
 
%% Reconstructed images from TestData
figure();
for i = 1:10
    subplot(4,5,i);
    imshow(xReconstructed(i))
end

But I got the following error

Error using imageDisplayValidateParams
Expected input number 1, I, to be one of these types:
 
double, single, uint8, uint16, uint32, uint64, int8, int16, int32, int64, logical
 
Instead its type was cell.
 
Error in images.internal.imageDisplayValidateParams (line 11)
validateattributes(common_args.CData, {'numeric','logical'},...
 
Error in images.internal.imageDisplayParseInputs (line 79)
common_args = images.internal.imageDisplayValidateParams(common_args);
 
Error in imshow (line 253)
    images.internal.imageDisplayParseInputs({'Parent','Border','Reduce'},preparsed_varargin{:});
 
Error in data_process1 (line 51)
    imshow(xReconstructed(i))

When I compare to the example,

https://www.mathworks.com/help/deeplearning/ref/trainautoencoder.html

the issue is with Train{i}and Test {i}, The model is not trained probably with given data. When I compared with example in MATLAB for digitTrainCellArrayData and digitTTestCellArrayData I am having problems and my cell array which are not the same as in example

My cell arrays for Train{i}and Test {i} are like this like for example Train{i}

But in the example in MATLAB for digits, digitTrainCellArrayData and digitTestCellArrayData are like this

digitTrainCellArrayData

In addition to that. Range in example is from 0 – 1 while my data is from 0 – 255

Kindly looking for your support

[drummer]

uHm, I see, so you should insert Walter's suggestion while you're transforming your imds.

for i = 1:numImages
    img = readimage(TrainData, i);
%     img2= imshow(img, 'InitialMagnification', 800);
    img3= im2double(imresize(img, [100 100])); % inserting Walter's suggestion.
    img4= imshow(img3, 'InitialMagnification', 800);
     drawnow;
    Train{i} = (img3);         %#ok<SAGROW>
 
end

Check if it works to get your image range.

[Walter Roberson]

https://www.mathworks.com/help/deeplearning/ref/autoencoder.predict.html

• If Xnew is a cell array of image data, then Y is also a cell array of image data, where each cell contains the data for a single image.

    Test{i} = (img6);      %#ok<SAGROW>

So Test is a cell array.

    xReconstructed = predict(autoenc,Test);

You are passing that cell array to predict() and so you are getting a cell array out of it.

        imshow(xReconstructed(i))

Using () indexing on a cell array gives you a cell array output. However, imshow() cannot display cell arrays. You need

       imshow(xReconstructed{i})

[Abdussalam Elhanashi]

Hi Walter

Thanks for your support

For Train and Test that i have is like this

But I want like this

How to make that to be as columns and raws

[Walter Roberson]

We already said to use im2double() to change the scale.

You might possibly also want to take im2double( uint8(255) - XTrain{1,1} ) if you want to change the white (255) into black (0)

[Abdussalam Elhanashi]

Hi walter

Thanks for your help

What i want is to have minimize the color for back ground of reconstructed images and here in the the currect results i have

Test images as input to detector

Reconstructed images as output from the detector

as you can notice that background is still with black color and it has to be white color the same as Test images or almost similar

[Walter Roberson]

as you can notice that background is still with black color

No, I do not notice that. The background looks white to me in the reconstructed images.

[Abdussalam Elhanashi]

Hi Walter

Yes, What I mean is it possible to increase brightness for background to be as original Test images or that is normal for Reconstructed images to be like this

[Walter Roberson]

reconstructed(imbinarize(reconstructed)) = 1;

[Abdussalam Elhanashi]

Hi Walter

I tried ths line but i got error

reconstructed(imbinarize(xReconstructed)) = 1;

Error using imbinarize
Expected I to be one of these types:

uint8, uint16, uint32, int8, int16, int32, single, double

Instead its type was cell.

Error in imbinarize>validateImage (line 262)
validateattributes(I,supportedClasses,supportedAttribs,mfilename,'I');

Error in imbinarize>parseInputs (line 198)
validateImage(I);

Error in imbinarize (line 134)
[I,isNumericThreshold,options] = parseInputs(I,varargin{:});

[Walter Roberson]

       reconstructed = xReconstructed{i};
       reconstructed(imbinarize(xReconstructed)) = 1;
       imshow(reconstructed)

[Abdussalam Elhanashi]

Hi walter I tried this but i got error

reconstructed(imbinarize(xReconstructed)) = 1;


Error using imbinarize
Expected I to be one of these types:

uint8, uint16, uint32, int8, int16, int32, single, double

Instead its type was cell.

Error in imbinarize>validateImage (line 262)
validateattributes(I,supportedClasses,supportedAttribs,mfilename,'I');

Error in imbinarize>parseInputs (line 198)
validateImage(I);

Error in imbinarize (line 134)
[I,isNumericThreshold,options] = parseInputs(I,varargin{:});

[Walter Roberson]

       reconstructed = xReconstructed{i};
       reconstructed(imbinarize(reconstructed)) = 1;
       imshow(reconstructed)