fast way to update image objects in Matlab figure

image-processing, matlab

Solution

If all of the images have same size, you can use the `image` command once, and then update the `CData` property of the axes.

videoObj = VideoReader('sample.avi');
nFrames = videoObj.NumberOfFrames;
mov(1:nFrames) = ...
    struct('cdata', zeros(vidHeight, vidWidth, 3, 'uint8'), 'colormap', []);
for k = 1 : nFrames
    mov(k).cdata = read(interObj, k);
end

tic
    image(mov(1).cdata);
    imageHandle = get(gca,'Children');
for i=1:number
    frame = mov(i).cdata;
    set(imageHandle ,'CData',frame);
    drawnow;
end
secPerFrame = toc/number

This will still be slow, since Matlab figures are not optimized for showing videos. If the application allows to do pre-processing, save it as a movie file to the disk, and call external program to show it.

Problem

I need to display updated images as fast as possible in a Matlab figure. Each image is processed and then displayed. However, the speed of displaying of color images is pretty slow. For example, when I run the following code ``` videoObj = VideoReader('sample.avi'); nFrames = videoObj.NumberOfFrames; h = videoObj.Height; w = videoObj.Width; mov(1:nFrames) = struct('cdata', zeros(h, w, 3, 'uint8'), 'colormap', []); for k = 1 : nFrames mov(k).cdata = read(interObj, k); end tic for i=1:nFrames frame = mov(i).cdata; image(frame); drawnow; end secPerFrame = toc/nFrames ``` it takes `secPerFrame = 0.012` seconds to update each frame. Where each frame is a 640x480 pixel RGB image. So, if I want to process a video stream at 30 frames per second, this leaves "only" `0.033 - 0.012 = 0.021` seconds for actual image processing after subtracting the overhead related to image display. Is there a faster way to update image objects in Matlab?

Original source