I am doing some staff with mean shift segmentation. I found the code called EDISON(http://coewww.rutgers.edu/riul/research/code/EDISON/) ,which has implemented a good one. Also, opencv has a function called cvPyrMeanShiftFiltering.
But they are different in the following picture. In the EDISON system, the two parameters are 7 & 6.5. In opencv, I use
cvPyrMeanShiftFiltering(result, result, 7, 6.5,level); (result is converted to LUV already) The borders are not clear there in the right image. I also tried level=0…
Can anyone help? I want to achieve the same result with the help of openCV. (I’ve implemented the clustering step already)

I tried another code from rsbweb…and the effect is much better than openCV!
Here is my new result
I changed some part of the code, using Luv rather than YIQ to gain better result. The color equation is from easyrgb
Here is my code