This is machine translation

Translated by Microsoft
Mouseover text to see original. Click the button below to return to the English verison of the page.

Note: This page has been translated by MathWorks. Please click here
To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.

Live Motion Detection Using Optical Flow

This example shows how to create a video algorithm to detect motion using optical flow technique.This example uses the Image Acquisition Toolbox™ System Object along with Computer Vision System Toolbox™ System objects.


This example streams images from an image acquisition device to detect motion in the live video. It uses the optical flow estimation technique to estimate the motion vectors in each frame of the live video sequence. Once the motion vectors are determined, we draw it over the moving objects in the video sequence.


Create the Video Device System object.

vidDevice = imaq.VideoDevice('winvideo', 1, 'MJPG_320x240', ...

                             'ROI', [1 1 320 240], ...
                             'ReturnedColorSpace', 'rgb', ...
                             'DeviceProperties.Brightness', 130, ...
                             'DeviceProperties.Sharpness', 220);

Create a System object to estimate direction and speed of object motion from one video frame to another using optical flow.

optical = vision.OpticalFlow( ...
    'OutputValue', 'Horizontal and vertical components in complex form');

Initialize the vector field lines.

maxWidth = imaqhwinfo(vidDevice,'MaxWidth');
maxHeight = imaqhwinfo(vidDevice,'MaxHeight');
shapes = vision.ShapeInserter;
shapes.Shape = 'Lines';
shapes.BorderColor = 'white';
r = 1:5:maxHeight;
c = 1:5:maxWidth;
[Y, X] = meshgrid(c,r);

Create VideoPlayer System objects to display the videos.

hVideoIn = vision.VideoPlayer;
hVideoIn.Name  = 'Original Video';
hVideoOut = vision.VideoPlayer;
hVideoOut.Name  = 'Motion Detected Video';

Stream Acquisition and Processing Loop

Create a processing loop to perform motion detection in the input video. This loop uses the System objects you instantiated above.

% Set up for stream
nFrames = 0;
while (nFrames<100)     % Process for the first 100 frames.
    % Acquire single frame from imaging device.
    rgbData = step(vidDevice);

    % Compute the optical flow for that particular frame.
    optFlow = step(optical,rgb2gray(rgbData));

    % Downsample optical flow field.
    optFlow_DS = optFlow(r, c);
    H = imag(optFlow_DS)*50;
    V = real(optFlow_DS)*50;

    % Draw lines on top of image
    lines = [Y(:)'; X(:)'; Y(:)'+V(:)'; X(:)'+H(:)'];
    rgb_Out = step(shapes, rgbData,  lines');

    % Send image data to video player
    % Display original video.
    step(hVideoIn, rgbData);
    % Display video along with motion vectors.
    step(hVideoOut, rgb_Out);

    % Increment frame count
    nFrames = nFrames + 1;


In the Motion Detected Video window, you can see that the example detected the motion of the notebook. The moving objects are represented using the vector field lines as seen in the video player.


Here you call the release method on the System objects to close any open files and devices.



Was this topic helpful?