Alert icon
We're changing our privacy policy. This stuff matters.  Learn more  Dismiss

Kalman Filter Assisted Object Tracking

Loading...

Sign in or sign up now!
Alert icon
Upgrade to the latest Flash Player for improved playback performance. Upgrade now or more info.
5,406
Loading...
Alert icon
Sign in or sign up now!
Alert icon

Uploaded by on Nov 12, 2008

To start with, I used the color thresholding method for originally the initial tracking task. For my test object, I used a small piece of red foam which I manipulated and moved around using a white stick. For the background I used a sheet of paper with multiple colors, patterns, and writing on it. I moved teh bakcground as the foam was moving to create noise and distortion to specifically test the Kalman filter.

I 'closed' the threshholded image first with a disk of radius 5 then immediatly 'opened' it with a disk of radius 19 to cut out any small objects that were kept by the thresholding to isolate the intended object.

From the blob analysis block, I created a '+' to track the actual centroid of the blob. I then inserted the Kalman filter block as another offshoot from teh centroid output of the blob analysis block. From there I took teh estimated and predicted outputs from the Kalman block and drew a 'x' for the prediction, and a 'o' for the estimation of the centroid.

Inside the Kalman Filter block, I played around with the Process Noise Covariance matrix and the Measurment Noise Covariance matrix. My final matrx values are as follows.

PNC - 10*eye(6)

MNC - [2 0; 0 3]

This shows that the model and noise is fairly low and close to the actual data. The estimation and prediction only need small corrections.

I was successful in predicting the blobs movement. The 'x' , for the most part, moves jsut slightly ahead of the actual centroid accuratly predicting its movement with little distortions. The estimation output by the Kalman filter is very accurate as well. The error is small, the 'o' stays very close to the actual centroid at all times.

Category:

Science & Technology

Tags:

License:

Standard YouTube License

Link to this comment:

Share to:

Uploader Comments (puregame13)

  • Hey, nice vid. What's the soundtrack? Did you use Matlab code or a Simulink block diagram approach to implement this?

  • @colmanian I used the Simulink Block Diagram with embedded matlab m-file code. the soundtrack is metallica

see all

All Comments (10)

Sign In or Sign Up now to post a comment!
  • Hi, great work, i like it

    can you send to me the source code

    my email is : sondabousnina@gmail.com

    thank you :)

  • What is your process model?

  • can you estimate the X,Y of center of moving object using simulink only

  • can i contact u via emel about this project? my emel is lia_safiyya@yahoo.com

  • Hey, you did a really nice job here. Can you recommend me some documentation? My objective is similar to yours. Any help is welcomed :) Keep up the good work :)

  • @puregame13 cool thanks

  • Hi..I am new to this..Can I know what is the purpose of using kalman filter..Is it to reduce error or predict the next location u r goin to move the object?

Loading...

Alert icon
0 / 00Unsaved Playlist Return to active list
    1. Your queue is empty. Add videos to your queue using this button:
      or sign in to load a different list.
    Loading...Loading...Saving...
    • Clear all videos from this list
    • Learn more