+++ /dev/null
- <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">\r
-<HTML>\r
- <HEAD>\r
- <TITLE>Condensation Tracker</TITLE>\r
- <!-- Changed by: Jean-Yves Bouguet, 31-Dec-1999 -->\r
- <!-- Formatted and proofed in GNU Emacs by: Michael Chu, 00.06.05 -->\r
- <META NAME="GENERATOR" CONTENT="Microsoft FrontPage 4.0">\r
- </HEAD>\r
- <BODY TEXT="#000000" BGCOLOR="#FFFFFF" LINK="#0000CC" VLINK="#0000FF" ALINK="#FF0000">\r
- <a name="top">\r
-\r
- <CENTER>\r
- <TABLE CELLSPACING=0 CELLPADDING=5 WIDTH="90%" BGCOLOR="#6A9BED" NOSAVE >\r
- <TR NOSAVE>\r
- <TD NOSAVE>\r
- <p align="center"><I><FONT SIZE=+4>Condensation Tracker\r
- </FONT></I></p>\r
- </TD>\r
- </TR>\r
- </TABLE>\r
- </CENTER>\r
-\r
-\r
- <p align="center">\r
-\r
-\r
- <BR>\r
-\r
- </a><img border="0" src="CondensationDemo.jpg" width="712" height="279">\r
- <a name="top">\r
-\r
- <BR><BR>\r
-\r
- <H1>Description</H1>\r
-\r
- </a>The library provides a Condensation engine for tracking purposes. In the\r
- application described below, the CAMSHIFT color tracking algorithm is used\r
- to provide measurements for the X-Y tracking of a face (or other colored\r
- object). The CAMSHIFT algorithm implemented here just uses a simple 1-D Hue\r
- histogram sampled from the object using an HSV color space. Therefore,\r
- the color tracker may need some tuning as described below (see CAMSHIFT demo\r
- description page for more details).\r
- <br><br>\r
- Note that this filter requires that you have DirectShow 6.0 or greater installed\r
- with <B>graphedt.exe </B>(DirectShow 6.0 for NT4.0, DirectShow 7.0 for\r
- Win 98 or Win2000)\r
- (<A HREF="http://www.microsoft.com/downloads/release.asp?ReleaseID=16927">http://www.microsoft.com/downloads/release.asp?ReleaseID=16927</A> ~128MB[!]).\r
- You will also need a video camera source filter which should\r
- be supplied by your USB camera vendor. <!-- except most don't support NT4.0. -->\r
-<!-- %%Some friendly advice from Uncle Gary?? -->\r
-<!-- Win2000 is your best bet for development, for just running the filter graph, Win98 will do.. An example filter graph is shown below (with the calibration filter processing a video source from a Matrox Meteor II card that we wrote ourselves) and then passing the output to a video renderer. -->\r
-\r
- Microsoft* Windows* 2000 may be the most suitable operating system\r
- for development since many vendors do not supply filters for\r
- Windows NT 4.0. However, Windows 98 will be adequate for just\r
- running the filter graph.\r
- <h1>Condensation Color Tracker Demo</h1>\r
- You should setup or launch the following DirectShow filtergraph (of\r
- course, your capture source will depend on your USB camera):<br><br>\r
- <center><img border="0" src="CondensationGraph.jpg" width="655" height="91"></center>\r
- <h2>Controls</h2>\r
- The control panel (obtained by right-clicking) for the\r
- Condensation filter in the filter graph above looks like:<br><br>\r
-\r
- <center><img border="0" src="CondensationDialog.jpg" width="386" height="280"></center>\r
- <br>\r
- <b><font size=+2>[1]</font></b> This control moves and\r
- sizes the\r
- sample window when you are not tracking.\r
-\r
- <br><b><font size=+2>[2]</font></b> This is the minimum saturation for a\r
- Hue pixel to be accepted threshold.\r
-\r
- <br><b><font size=+2>[3]</font></b> This is the minimum brightness for a\r
- Hue pixel to be accepted threshold.\r
-\r
- <br><b><font size=+2>[4]</font></b> This allows you to toggle between the\r
- raw video image and the Hue histogram backprojected image.\r
-\r
- <br><b><font size=+2>[5]</font></b> Samples color under box [1] and starts Condensation\r
- tracking.\r
-\r
- <br><b><font size=+2>[6]</font></b> Stop Condensation tracking.\r
-\r
-\r
- <h2>How to...</h2>\r
- <h3>Start Tracking</h3>\r
- 1. Select Normal view [4].<br>\r
- 2. Set the thresholds to zero [2] & [3].<br>\r
- 3. Move the sample box using [1] so that it covers the colored object to be tracked.<br>\r
- 4. Press "Start Track" [5].<br>\r
-\r
- <h3>Tuning the tracking:</h3>\r
- 1.Select Back Project view [4]. The goal is to have\r
- the image such as on the left below to look like the cleaner image on\r
- right below.<br>\r
- <br>\r
- <center><img border="0" src="../Kalman/kalmanbackNClean.jpg" width="649" height="240"></center>\r
- <br>\r
- 2. Increase minimum saturation threshold [2] to clean up\r
- the image without diminishing the object of interest too much.<br>\r
- 3. Increase minimum brightness threshold [3] to finish the\r
- "clean up" job. Don't turn too high or else bright spots\r
- on the tracked object will become "holes".<br>\r
- 4. Turn normal view back on [4].<br>\r
- <h2>Display</h2>\r
- The meaning of the normal view display is as follows with\r
- reference to the figure below:<br><br>\r
- <center><img border="0" src="../Kalman/kalmanTrackShot.jpg" width="320" height="240"></center>\r
- <font size=+2><b>[A]</b></font> Velocity direction and\r
- magnitude estimation.<br>\r
- <font size=+2><b>[B]</b></font> Sample window if not\r
- tracking, tracking window if tracking.<br>\r
- <font size=+2><b>[C]</b></font> Estimated location of \r
- the object.\r
- <br>\r
- <br>\r
- <hr noshade size=2 width=100%>\r
-\r
- </BODY> </HTML>\r
projects / opencv / blobdiff