Plot sequence of images on top of a trajectory

I have a a 3D array G that has frames of a ball moving. I also have a variable named traj, that gives me the coordinates of the trajectory of the ball (traj.x and traj.y ). The arrays are attached.

I need to plot the frames on top of the trajectory. As if the trajectory is the background of the frames.

If you use sliceViewer(G) you will see the movement. I just want the trajectory to be the background of this movement.

I don't know even how to start.

Any ideas?

Answer 1

[DGM]

Here's a start. Note that the alignment is just guesswork on my part. Also, I don't know why there's a frame offset between the image and the xy data.

load G.mat
load traj.mat

alph = 0.5; % scalar FG opacity
nframes = 100;
frameoffset = 2; % the image appears to be 2 frames ahead of the plot data?

% set up the figure
hp = plot(traj.x,traj.y,'-o'); hold on
hi = image(uint8(G(:,:,1))); hold on % need to reassert hold
hi.XData = [-92 85.6]; % these are estimates
hi.YData = [81.1 -87];
xlim([-50 50])
ylim([-50 50])
set(gca,'ydir','normal','dataaspectratio',[1 1 1])
colormap(gray(256))
grid on

% run the loop
for f = 1+frameoffset:nframes
    % update the plot data
    hp.XData = traj.x(1:f);
    hp.YData = traj.y(1:f);
    
    % update the image data and reassert
    hi.CData = uint8(G(:,:,f-frameoffset));
    hi.AlphaData = (G(:,:,f-frameoffset)<128)*alph;
    
    drawnow();
    pause(0.1) % wait
end

If you don't have some sort of spatial calibration data, I can show you how I got the estimates that I used. It'd probably be best if you didn't have to resort to doing what I did.

[L]

Hi @DGM, thanks so much.

I don't have any clibration data. Can you walk me through the estimates?

[DGM]

This is how I did it:

load G.mat
load traj.mat

nframes = 100;

% get centroid positions for all frames
xy = zeros(nframes,2);
for f = 1:nframes
    S = regionprops(G(:,:,f)<128,'centroid');
    xy(f,:) = S.Centroid;
end

% select only the segments where the two sequences correspond
xy0 = [traj.x; traj.y].';
xy0t = xy0(3:end,:); % trimmed traj data
xyt = xy(1:end-2,:); % trimmed centroid data

% get the fit parameters
f1 = fit(xyt(:,1),xy0t(:,1),'poly1')

f1 = 
     Linear model Poly1:
     f1(x) = p1*x + p2
     Coefficients (with 95% confidence bounds):
       p1 =       3.626  (3.612, 3.641)
       p2 =      -95.73  (-96.13, -95.32)

f2 = fit(xyt(:,2),xy0t(:,2),'poly1')

f2 = 
     Linear model Poly1:
     f2(x) = p1*x + p2
     Coefficients (with 95% confidence bounds):
       p1 =      -3.431  (-3.444, -3.418)
       p2 =       84.54  (84.19, 84.88)

% these are the ranges used for hi.XData and hi.YData
xdrange = [1 size(G,2)]*f1.p1 + f1.p2

xdrange = 1x2

-92.0992 85.5811

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

ydrange = [1 size(G,1)]*f2.p1 + f2.p2

ydrange = 1x2

81.1059 -87.0186

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

If you don't have CFT, you can just use polyfit() instead of fit():

f1 = polyfit(xyt(:,1),xy0t(:,1),1);
f2 = polyfit(xyt(:,2),xy0t(:,2),1);
xdrange = [1 size(G,2)]*f1(1) + f1(2)

xdrange = 1x2

-92.0992 85.5811

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

ydrange = [1 size(G,1)]*f2(1) + f2(2)

ydrange = 1x2

81.1059 -87.0186

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>