Maybe you could give it a try with this open source tool http://www.kinovea.org/ 

Based on the shape of your particles it is difficult to say how to measure their centers. I think it is up to you to decide what should be considered a center based on physical properties of the "particles". For example, you might want to consider the coordinates of the small "dot" below each (larger) "blob", or you may want to consider the center of mass of the "blob", or you may want to consider fitting a circle to the "blob" and small "dot" hanging below the blob. If the "dot" is connected rigidly to the "blob" - it does not matter whose coordinates you are computing (unless there are rotations).

Once you know the coordinates, it is very simple to exclude moving particles: simply check that in the second image there is a particle within a small radius from the coordinates of the same stationary particle in the first (reference) image.

In fact, if you do not need to know the coordinates of the particles, and just want to get an image of stationary particles only, then I would suggest to simply overlap the two images (assuming the camera did not move) and create a mask using the following formula:

mask=If Abs(reference-image) < small_tolerance then 1 else 0

and then multiply the image by the mask. This should remove moving particles.

Maybe you could give it a try with this open source tool http://www.kinovea.org/ 

Thanks all.

if the particles have differnet shapes, you can use a shape descriptor to describe each particle and record their corresponding locations (x, y) for each frame. then you can compare the records for each frame and remove any particle whose location has changed.

There is a method of Digital Image Correlation. I do not know about any free software.  However, the idea may be implemented in any high level software package Matlab/OctaveScillab// Python.

http://en.wikipedia.org/wiki/Digital_image_correlation

https://www.researchgate.net/publication/264277037_3D_Digital_Image_Correlation_for_tracking_displacements_of_canvas_paintings_with_natural_texture

Conference Paper 3D Digital Image Correlation for tracking displacements of c...

Dear Reza,

Do you want to remove the whole moving particles and just stay fixed particles?

Dear Abbas Oladi

all my particles are circle (2D sphere)

Do you have any code or software?

Dear Alireza

Yes, i want to remove all moving particles and detect location of fixed particles

I dont have a code but I think a matlab algorithm can be developed very easily. You just need to establish a way of differentiating one particle from the other. If they are all 2D Spheres then the diameters or colour should be adequate. 

Dear Reza,

You can use following project.

If you have problems, Say to me.

Hello:

I developed a very robust algorithm to track moving white blood cells, it may be possible to use it to track your spheres, the algorithm is based on Matlab, it is released open-source and there is a comprehensive user manual for it, give it a try:

www.phagosight.org

http://www.phagosight.org/

Rather than tracking the particle, you can simply monitor the locations as you move from frame to frame. Any location that appear in all the frame should be kept.

Your images look like excellent candidates for the feature detection and tracking algorithms described by John Crocker and Eric Weeks at

http://www.physics.emory.edu/faculty/weeks//idl/

Versions of this code are available in matlab, python and IDL, among many other languages.  The suggested link includes instructions on how to track colloidal particles in images very similar to yours.

Thank you all

Alireza Rahmani : thanks in advance

Dear Reza,

Why don't you do any image processing before to keep only static particles ? You could average a certain amount of frame and process to a binarization of the level intensity to keep the brightest particles.

Regarding to Matlab you could use the 2D cross correlation  if your particles have a sufficient large shape, or maybe a centroid detection. with the function "regionprops"  :

http://fr.mathworks.com/help/images/ref/regionprops.html.

In the last case you can use the PTVlab toolbox, wich is a freeware toolbox on matlab to track particles :

http://ptvlab.blogspot.fr/

Best regards,

Dear Raphael

Thanks

hello

regarding your problem i suggest you use particle swarm optimization method for multi-object tracking implement it using matlab it is fairly fast enough to crunch all your data focus your search area on the above keywords hope you get your answers there

best regards

hi, I think  particle swarm optimization method is helpful for u r work and also look into the below link

http://site.physics.georgetown.edu/matlab/