Image Segmentation for object detection

[Sammura]

Hi Mr.cesar
I'm performing image segmentation on images that contain more than one or two objects , my goal is to get every object in a single image .
I used color filtering with the help of Aforge library " DifferenceEdgeDetector filter " then I applied a threshold function to get the edges without details
then I applied "CollectActivePixels function " to get the pixels which are the edges ,,, and here's my problem !! how could I get the connected shapes that form an object then I can cut it from the original colored image according to the saved coordinates which I have in the image matrix
I'm facing a speed and bad performance problem because I can't use a slide window on the image !
any suggestion ?
I've attached a photo - the debug ...

best regrads
Sammura

[César]

Hi Sammura,

This question seems more specific to AForge.NET, but in any case, you could attempt doing successive Closing operations using AForge.NET's Closing class. Please take a look on this link for the documentation, it gives an example on how it can "close" those gaps in your image.

After you have obtained more smooth blobs, you can use AForge.NET's blob extraction processing routines. Please take a look on this link on how you could extract all blobs from an image.

I hope it helps!

Best regards,

Cesar

[Sammura]

Hi Mr.cesar 
thank you for your help ,i'll try this step 
and hope it work as i want 

best regrads
Sammura 

[shailaja...@gmail.com]

Hi,

I am newbie to Accord.net framework.
I am also performing image segmentation on image that contain more than one or two objects , my goal is to get count of objects in that image.

So can somenone share me piece of code for this scenario.

Thanks in advance.

Regards,
Shailaja