Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

mean shift clustering

3 views
Skip to first unread message

Deepak Joshi

unread,
Apr 26, 2021, 7:54:22 AM4/26/21
to
Use 'flower.png' and 'parrot.png' as two different inputs.
In this case, the points are in 5D space (x,y,R,G,B).
Use a Gaussian kernel for mean shift procedure with spatial standard deviation σsp and intensity standard deviation σint. Kindly tune and notify σsp∈[9,12] and σint∈[18,25] accordingly.
While performing the mean shift iterations on any point, compute the weighted average using only those points falling within a 23×23 square centered at (xt,yt) where (xt,yt) are the spatial coordinates of the point in the current iteration (note that these spatial coordinates will change in every iteration).
Display the initial image, and the filtered image.
Given the filtered image, we will now perform segmentation on it using K-means in the joint spatial and intensity space, using K=100.
Display the segmented image - assign all pixels in a segment a color value equal to the average of the color of all pixels in that segment.

Can anyone post the code for this?
very urgent
0 new messages