05 July 2007

The current application hamster detection is based on white color detection and histogram detection. I have temporally removed the histogram detection from the application.

To reduce the interference caused by shininess of the side panels we have panted them with mat black color. And it worked very well.

Current area detection is based on selecting any area bigger than 300. This detection is too simple specially if there is interference. I have modified this to select a range.

Several other functions such as compactness, roundness can be introduced to create more stable algorithm. But introduction of these functions will take up lots of computations. The current algorithm seems to be sufficient at this point.

I have tested tracking application with the large white hamster and the results are quite good. But the hamster can move under the side panels and it can not be seen by the camera.

Lighting is very critical to tracking applications. The setup can not depend on external lighting. The system should have an independent lighting system.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-Share Alike 2.5 License.