Collecting Image Dataset into a variable dimension mismatch problem

Supoosed i Have declare a Maxtrix variable

train_x = zeros(64,128,3,15);

The train_x variable is 4-D double. which store up to 15 images it is a image dataset

I have an Images whose dimensions are 64 x 128 x 3 uint8 and 64 x 128 uint8

Res_none = imresize(Img_none, [64 128]);   % -> where Res_none is 64 x 128 uint8

test_x(:,:,:,1) = Res_none; % <- due to insufficient dimension

I am trying to store an image into the train_x 4-D array but because images is 64 x 128 uint8, but it is insufficient dimension to store because it only stores an image whose dimensions are 64 x 128 x 3 uint8.

Is it possible that I can store that image 64 x 128 uint8 to the train_x (dataset) so that I can retrieve that same file from train_x (dataset)?

[jgg]

Why not just declare train_x = zeros(64,128,8,15) instead?

I don't think you're going to be able to store a 64x128 image with 8 bits of information per pixel into a 62x128x3 matrix without collapsing the 8 bits down.

Since 8 bits is only 2^8, could you not convert it to an integer, then store the integer instead of storing it bitwise?

Answer 1

[Dima Lisin]

If you are trying to store 15 RGB images of size 64 x 128 x 3, the tran_x should have the size [64,128,3,15], exactly like you have it. You may also want to make it a uint8 array from the start, like this:

train_x = zeros([64,128,3,15], 'uint8');

If you want to store 15 grayscale images of size 64 x 128, then train_x should have the size of [64, 128, 15].

If some of your images are RGB and some are grayscale, then there are several things you can do. You could make all images RGB.

imgRGB = cat(3, imgGray, imgGray, imgGray);

Then all of your images will have the size 64x128x3. Alternatively, you can use a cell array to store the images. Elements of a cell array can have different sizes and even data types.

[ken ken]

my imgGray is 64 x 128 unint8

after using this commands

imgRGB = cat(3, imgGray, imgGray, imgGray);

the result of the imgRGB value is 64 x 128 x 9 unint8

how may i make it to 64 x 128 x 3 unint8 so that i am able put into the train_x dataset

[Dima Lisin]

Wait, if your result is 64 x 128 x 9, then imgGray must be 64 x 128 x 3.