If you could post an example image, I'd be happy to show how our new software MIPAR at http://MIPAR.us could very likely do the trick!

Cheers,

John

Hi,

You can use imageJ to do that, "Analyse Particles" is not disiged to detect circular objects...

You have to play with the size and the circularity parameters...

Can you show us one of your picture ? So that, I could give you some more precise advices...

Don't hesitate to check the documentation :

https://imagej.nih.gov/ij/docs/guide/146-30.html#toc-Subsection-30.2

Regards,

Arthur

Hi Mr.Soon Hwee Ng,

Can you please share one sample image, we will try to solve?? I have the following doubts.

1. Whether the image consists of only rectangles/squares? Any other objects will be there with some other shapes??

2. What about the size of these objects? All have the same size or not??

Problem is interesting, if you clarify this, we will solve it.

 Hi all,

Thanks for your responses, and sorry for the long silence. I had thought that I would get an email notification when there is a response. When I didn't receive any email, I thought there was no answer. 

So I attached an image. The squares/rectangles are actually pulses of a song produced by a cricket. They might be irregularly shaped and not the same size, and also subject to how much I zoom in to look at. 

Thanks again!

Hello Mr.Soon Hwee Ng,

Since your input is a signal, why dont you use some signal processing stuffs to count the pulses? Why are you using image of a signal? Your problem of counting can be easily solved in signal domain. 

In case of image, Convert the image into binary and then apply morphological opening to the image. All the connections between pulses will go. Then you just have to count the objects. Finding the type and size of morphological structuring element is very important.

Thanks for triggering my brain.

Regards,

Prabhakaran S.

Hi Prabhakaran,

Thanks for the reply. I'm using image as we also need to sample the duration of pulses, the interval between pulses, etc. 

I'm not sure how to use signal domain or morphological opening you mentioned. I'm a biologist, so can't really understands those jargons.

Thanks,

Soon

Hi Soon,

In ImageJ, just look for binary operations. I think you can find Morphological opening. In image processing, we use this to isolate touching objects in our binary image. In your case all the pulses will be isolated. Before this you have to convert your image to binary format( black and white image ).  This conversion is explained in the link as Thresholding. Then try your 'Analyze Particles' to count them.

Good luck.

Regards,

Prabhakaran S.

https://imagej.nih.gov/ij/docs/pdfs/examples.pdf

Hi Soon,

Sorry for the delay here.I was able to count the pulses very quickly and completely automatically in MIPAR (available at http://MIPAR.us). I have attached the result of the auto pulse-finding (each pulse has been randomly colored to show it is separate. I have also attached the MIPAR recipe file that you can just run in the Image Processor (or the Batch Processor to count pulses from multiple images) and it will do the pulse finding and counting for you. In this image, 57 pulses were counted. Please let me know of any questions!

Cheers,

John