How can i replace this code to fast up?

 for i=1:500,
    subplot(23,23,i);
    x=vishid(:,i);
    imagesc(reshape(x,imagesize));
    colormap gray;
end

Here,vishid is 784*500 matrix, imagesize=28*28

[Geoff Hayes]

Subha - the above code is trying to create 500 subplots within the figure, so it may take some time! :) Adding a drawnow command after the colormap gray line shows how the first 30-40 subplots get drawn relatively quickly, but over time, it takes longer and longer to add the newest subplot to the figure. Even just doing the above with something as simple as

 for k=1:500
    h = subplot(23,23,k);
    set(gca,'XLim',[0 1]);
    drawnow;
end

takes a long time (this was just to see if the bottleneck had to do with the reshape and imagesc - not the case).

Is there a need to show all 500 subplots on the same figure, or could you break it into two figures of 250?

[subha]

Thanks for your post. Its not necessary to show all 500 plots in a same figure, But it will be better, if it is in same.

If time reduce by implementing this in more than one figure, i am happy about it

Answer 1

[Jan]

Do you really need 500 images with separate axes objects? You can draw 500 im ages in one axes, when wet the coordinates appropriately. This would be much faster.

[subha]

How?

[Image Analyst]

I totally agree. 500 subplots is absurd. Most of the space will be taken up be white space and you won't see the images.

[Matt J]

Whether it's worthwhile depends on how well you need to see the images. Thumbnail views have their uses.

Answer 2

[Matt J]

Making the figure invisible until all subplots have been established might help,

   h=figure; 
   set(h,'Visible','off')

    for i=1:500,
        subplot(23,23,i);
        x=vishid(:,i);
        imagesc(reshape(x,imagesize));
        colormap gray;
    end

   set(h,'Visible','on')

[subha]

Thanks.Actually, in my case, already my figure appears only after it plots all the 500 plots. so, i am not sure, this will help. Let me try again and come back to u

[subha]

I have done . But no difference with speed

Answer 3

[Image Analyst]

If you have enough memory, use the montage() function. Or else just use regular indexing. Are the images grayscale or color?

[subha]

They are greyscale

[Image Analyst]

OK. So did you try what I suggested? What happened?

[subha]

I go with regular indexing method first.. can you please help me to understand and use regular indexing to plot these values.

I want to plot an image. So i used imagesc func, want to plot in grayscale, so used colormap. since i have a matrix, i have to use for loop. i have to take column wise , so used inc= 1:c, i need them to appear in 28 *28 pixel, so reshaped each column into 28*28. next my aim is i want plot this each column as one image. so i will have no of images = no of columns. i have to plot each column as one element of matrix. so, assigned k(inc) as an image. each k(inc ) will have one image. but result is not as i asked in this question.. just showing some randomn pixels..

could u please help to solve this.

function []= imageout(x);
[r,c]=size(x);
for inc=1:c,
  l= x(:,inc);
  l= reshape(l, [28,28]);
  k(i)=imagesc(k(i),[0,1]);

 end

colormap(gray); end

[Image Analyst]

Seems like a weird thing to do, but I did it. Run the mfile attached below the image. I'll say one thing, it creates some interesting animation as it scans along the columns. Here is the final image it ends up with:

[subha]

thanks a lot for your interesting animation ..

[Image Analyst]

Does it do what you want? I thought it did what you said to do. If it does, please mark it "Accepted".