How to convert 1 column vector text file to rgb image?

I have a text file(.txt), how can I convert the 1 column vector (1x150528) to image 224x224x3 as shown below??

can show me the coding ?? thanks for helping me T.T

[Adam]

It depends how the column is ordered with respect to the image dimensions. A .txt file is a very odd way to store image data!

doc reshape

should help you with that though. Possibly just the most simple version

reshape( vec, [224 224 3] )

but as I say, it depends what order that column is in.

[Tham Wei Jian]

Here is my coding and output. Do you have method help me remove the line inside the image?This very annoying...

Please help me edit my coding...

============================================================

fid = fopen('input_data.txt','r');

img = fscanf(fid,'%x');

fclose(fid);

outImg = reshape(img,[224*3 224*1]);

outImg = outImg';

imshow(outImg,[])

=========================================================

[Geoff Hayes]

Tham - since your image is 224x244x3, then why are you doing

outImg = reshape(img,[224*3 224*1]);

since this will create a 672x224 (grayscale) image? Why not try what Adam suggested with

outImg = reshape(img, [224 224 3]);

Also, why the %x

img = fscanf(fid,'%x');

and not %d?

Answer 1

[Athul Prakash]

Hi Tham,

As mentioned here already, your shape of image should be 224x224x3, and you should be using a 3 dimensional array.

Why Lines are there: When it's being reshaped to 672x224 instead, all 3 color channels are forced into the same dimension: The RGB values for each pixel now appear as different pixels along the same column. Thus, the image would have horizontal stripes, each stripe being the value of a different color channel. After tansposing, the lines become vertical lines ofcourse.

To show the color image:

Load using %d instead of %x.

fid = fopen('input_data.txt','r');
img = fscanf(fid,'%d');
fclose(fid);

Reshape into 3x224x224. The order in which the data was stored is important here. In the file given, it seems that channels are stored along the first dimension.

outputImg = reshape(img, [3 224 224]);

To display the image, we need to make sure that color channels are present along the last dimension, i.e. 224x224x3. Also, we need to transpose the image as well. We can reorder the dimensions of the image using permute()

outputImg = permute(outputImg, [3 2 1]);

See this documentation:

https://www.mathworks.com/help/matlab/ref/permute.html

Convert the image to type uint8 (very important) and call imshow()

outputImg = uint8(outputImg);
imshow(outputImg);