How to use estimateFlow between two images that are not video frames

50 views (last 30 days)
I am interested in computing optical flow between images that are not part of a video sequence using the built-in MATLAB function estimateFlow. That is, I would like to compute optical flow between two images in my workspace, Img1 and Img2. How can I do this?
Currently, I am combining the images into an array and running a for loop to obtain the optical flow:
ImgStack(:,:,1) = Img1; % combine the images to a sequence
ImgStack(:,:,2) = Img2;
reset(opticFlow); % clear previous optical flow
opticFlow = opticalFlowLKDoG; % At this point, I don't realy care what method is used
for t = 1:2
flow = estimateFlow(opticFlow,ImgStack(:,:,t));
end
This procedure is inelegant and quite cumbersome. I am doing this on a large number of images and I would like a more efficient way to do this. It would be nice if I could just do something like flow = estimateFlow(opticFlow,[Img1,Img2]).
Dose anyone know how to do this?
  2 Comments
Image Analyst
Image Analyst on 7 Aug 2020
What do you have to start with? A bunch of arrays called Img1, Img2, Img3, etc.? Or one array ImgStack?
chris crowley
chris crowley on 8 Aug 2020
This operation is nested inside of another operation within the code that only generates two images at a time. Img1 and Img2 are grayscale images of the same size. I could rewrite the code to produce an array that is all the Img1s and Img2s appended in an interlaced way into one very long array. This will limit how many images I can analyze because of memory constraints.

Sign in to comment.

Accepted Answer

Divya Gaddipati
Divya Gaddipati on 11 Aug 2020
You can use opticalFlow function for this purpose.
img1 = im2double(Img1); % the images should be in double
img2 = im2double(Img2);
opflow = opticalFlow(img1, img2);
h = figure;
movegui(h);
hViewPanel = uipanel(h,'Position',[0 0 1 1],'Title','Plot of Optical Flow Vectors');
hPlot = axes(hViewPanel);
imshow(img1)
hold on
plot(opflow,'DecimationFactor',[5 5],'ScaleFactor',60,'Parent',hPlot);
hold off
For more information, refer to the documentation on opticalFlow
  3 Comments
Walter Roberson
Walter Roberson on 8 Nov 2022
It depends exactly what is in the images. The images just might happen to be sensor readings from a flow meter that is directly estimating velocity at each point, perhaps using sonar or polarization techniques. The image files would have to be TIFF in order to store negative values properly...
So the call here is not definitely wrong... just likely wrong.

Sign in to comment.

More Answers (0)

Categories

Find more on Image Processing and Computer Vision in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!