CREU Project Blog

This is one of the earlier demo we wrote to experiment with blob detection.  The program uses the camera on the macbook or the Logitech webcam to capture images and track color blobs specified by the user.  After the program started, user can click on the color they want to track in the application window and the program will track blobs of that color and calculate the size of the blobs.

Blob tracking might at first seems like a trivial task; it was a lot more difficult than we previously anticipated.  The program picks up noises from the environment and it returns irrelevant blobs from time to time.  We try to overcome this by smoothing and performing segmentation on the image prior to searching for the color blobs.  The program is still prone to error if the color is too similar to the background color, the blob is too small or the lighting condition changes.  The source code can be downloaded here .  (Note: wordpress doesn’t allow user to upload zip file, so I renamed it to BlobDetectionDemo.pdf.  To decompress the source code, you have to change the file name to BlobDetectionDemo.zip.)

Figure 1. A screen shot of the program tracking a color finger wrap.  It puts a red box around of finger wrap and provide a label indicating its size.

Figure 2. We find that the size label increases quadratically as we move the finger closer to the camera.

We presented our project’s poster at the meetings of the mind hosted by Carnegie Mellon University’s Undergraduate Research Organization.  Here is the pdf for the poster.