dLife Vision Example
public class CokeCanTracker extends Controller {
private BlobFilter blobs;
private ImageSequenceCamera myCam;
public void startUp(Robot myRobot) throws VideoSourceCreationFailedException {
myCam = new ImageSequenceCamera();
myRobot.addDevice(myCam);
// Gaussian blur with width 7 pixels and standard deviation of 3
GaussianFilter gf1 = new GaussianFilter(7, 3.0);
myCam.addFilter(gf1);
// Match the colors found in the Coke can and turn them yellow.
MatchFilter mf2 = new MatchFilter(77, 186, 17, 79, 18, 83);
myCam.addFilter(mf2);
// Find at most 1 blob of at least 100 yellow pixels.
blobs = new BlobFilter(new Color(255, 255, 0), 100, 1);
myCam.addFilter(blobs);
}
public void step() {
// Draw a cyan circle to indicate the location of the Coke can.
Collection<Blob> blobCol = blobs.getBlobs();
Iterator<Blob> it = blobCol.iterator();
if (it.hasNext()) {
Blob b = it.next();
myCam.clearDrawing(Camera.RAW_VIDEO);
myCam.drawCircle(Camera.RAW_VIDEO,
(int)b.getCenterOfMass().x,
(int)b.getCenterOfMass().y,
10, Color.CYAN, true);
}
}
}
The example below, for use with the ControlCenter, uses blurring, color matching and blobification to track the location of a Coke can in sequence of images. In order to be used with the ControlCenter this class extends dlife.robot.Controller and provides two methods:
-
•startUp: This method is executed once when the ControlCenter’s Start Up button is clicked. It creates a camera and adds it to a robot. It also creates the necessary filters and adds them to the camera.
-
•step: This method is executed repeatedly by the ControlCenter after the Run button is clicked. It accesses data from the BlobFilter and draws a cyan circle to indicate the location of the Coke can.
Similar code could be used to perform image processing in a stand-alone Java program.