Third Year Project

The third year part of the project has been successful to the extent that the camera has been mounted on the mounting rig and is controllable on two dimensional axes via the serial port of a PC using a microcontroller (an Atmel AT89C52) to directly translate control characters into movement signals. This is controlled by a graphical user interface coded in Java.

Both of these components (the graphical user interface and the microcontroller circuit) were successfully designed and coded as part of the third year project.

Access to the video feed has not been achieved as part of the third year project due to problems discussed later.

for (i = 0; i < NUM_IMAGES; i++)
{
	images [i] = toolkit.getImage(imgname[i]);
	tracker.addImage(images[i],0);

	try
	{
		tracker.waitForAll();
	}
	catch (InterruptedException exc) {}

	while (!tracker.checkAll())
	{
		System.out.println("Please wait.......");
	}

	if ((images[i].getWidth(this) > IMAGE_WIDTH) 
				|| (images[i].getHeight(this) > IMAGE_HEIGHT))
	{
		images[i] = images[i].getScaledInstance(IMAGE_WIDTH,IMAGE_HEIGHT,0);
	}
}

protected boolean handleSinglePixel(int pixel)
{
	alpha = (pixel >> 24) & 0xff;
	red   = (pixel >> 16) & 0xff;
	green = (pixel >>  8) & 0xff;
	blue  = (pixel      ) & 0xff;
 
	redDifference = (redStored - red) * (redStored - red);
	blueDifference = (blueStored - blue) * (blueStored - blue);
	greenDifference = (greenStored - green) * (greenStored - green);
 
	totalDifference = redDifference + blueDifference + greenDifference;
	colourDifference = Math.pow(totalDifference,0.5);
 
	redStored = red;
	greenStored = green;
	blueStored = blue;

	if (colourDifference > IPthreshold)
	{
		return true;
	}
 
	return false;
}

public void findEdges()
{
	for (j = 0; j < bufferedImage.getHeight(); j++)
	{
		for (i = 0; i < bufferedImage.getWidth(); i++)
		{
			if ((handleSinglePixel(bufferedImage.getRGB(i,j)) == true) 
								&& (i != 0))
			{
				bufferedImage.setRGB(i,j,5435934);
				blackBufferedImage.setRGB(i,j,5435934);
			}
		}
	}

	for (j = 0; j < bufferedImage.getWidth(); j++)
	{
		for (i = 0; i < bufferedImage.getHeight(); i++)
		{
			if ((handleSinglePixel(bufferedImage.getRGB(j,i)) == true) 
								&& (j != 0))
			{
				bufferedImage.setRGB(j,i,5435934);
				blackBufferedImage.setRGB(j,i,5435934);
			}
		}
	}

	repaint();
}

public Circle trackMotion(Image image, Circle circle)
{
	// Do a dummy run to set the variables correctly
	handleSinglePixel(tempBufferedImage.getRGB(circle.getCentreX(),circle.getCentreY()));

	while (timeout < ((int) ((circle.getDiameter() / 2) * 1.1)))
	{
		if (handleSinglePixel(tempBufferedImage.getRGB(circle.getCentreX()
						+ timeout,circle.getCentreY())) == 1)
		{
			tempBufferedImage.setRGB(circle.getCentreX() + timeout,
							circle.getCentreY(),5435934);
		}

		timeout++;
	}

	while (timeout < ((int) ((circle.getDiameter() / 2) * 1.1)))
	{
		if (handleSinglePixel(tempBufferedImage.getRGB(circle.getCentreX()
						+ timeout,circle.getCentreY())) == 1)
		{
			tempBufferedImage.setRGB(circle.getCentreX()
						+ timeout,circle.getCentreY(),5435934);
		}

		timeout--;
	}

	// once the four edges have been found, do maths and create 
	//  a circle to return with the location of the position
	//   of the circle in it

	repaint();

	return paintedCircle;
}

private void captureMedia()
{ 
	Format formats []; 

	String nameCaptureDeviceVideo = "vfw:Philips VGA Digital Camera (Vesta):0";

	CaptureDeviceInfo deviceInfo = 
			CaptureDeviceManager.getDevice(nameCaptureDeviceVideo);

	VideoFormat videoFormat = null;
	formats = deviceInfo.getFormats();

	videoFormat = (VideoFormat) formats[1];

	dataSource = JMFUtils.createCaptureDataSource (null, null, 
					nameCaptureDeviceVideo, videoFormat); 

	if (dataSource == null)
	{ 
		JOptionPane.showMessageDialog(this, "1: Could not connect 
			to camera\nNot Installed or not plugged in", "Camera 
				Error", JOptionPane.ERROR_MESSAGE); 

		System.exit(1); 
	} 
	else
	{ 
		FormatControl control = ((CaptureDevice)dataSource).getFormatControls()[0]; 

		videoFormat = getDesiredFormat(control); 
		imageTransfer = new ImageTransfer(videoFormat); 
		control.setFormat(videoFormat); 

		try
		{ 
			dataSource.connect(); 
		} 
		catch (IOException ioe)
		{ 
			JOptionPane.showMessageDialog(this, "2: Could not connect to 
				camera\n"+ioe, "Camera Error", JOptionPane.ERROR_MESSAGE); 

			System.exit(1); 
		} 

		open(dataSource); 
	}
 
	pack(); 
	setVisible(true); 
}

stream.read(buffer); 
byte[] data = (byte[])buffer.getData();

public void close()
{
	if (mediaPlayer == null)
		return;

	try
	{ 
		dataSource.stop(); 
	}
	catch (IOException ee)
	{
		System.out.println("ERROR - Could not close capture device");
	}

	dataSource.disconnect(); 
	dataSource = null;

	mediaPlayer.stopAndDeallocate();
	mediaPlayer = null;
}

Execution Time Requirements

Analysis was done during a typical execution of the application using a commercial Java Profiler called "JProbe". This reveals internal information about Java programs such as method calls and method time details. The results of analysis on the main methods involved with the edge detection and motion tracking algorithms are show in Table 3.6a.

Method Name	Number Of Method Calls	Cumulative Method Time (%)
ImagePane.handleSinglePixel()	38624	0.1
BufferedImage.getRGB()	38624	0.7
IPSetup.findEdges()	1	0.8
IPMotion.trackMotion()	14	1.3

Table 3.6a : results of analysis of the application using JProbe

As can be seen from the results of the analysis, the motion tracking algorithm is the most resource-hungry. This could be expected due to the nature of what it is actually doing. The surprising outcome of this analysis is that the handleSinglePixel() method of the ImagePane class only takes up 0.1% of the total time of execution, even though it is called nearly forty-thousand times during execution.

This analysis shows that the algorithms have been designed well for speed and work effectively during execution without taking up too many system resources.