Fluorescent particle intensity measurment

[Hui]

Dear all,

I have some problems in measuring the intensities of fluorescent particles. The particles are monodisperse and are captured by a color camera. Because of the focus and temperature filed, in the image the particles showing different intensity (some are bright while some are dim) and outlines (some are sharp as they are in focus and some are blurry because of out-of-focus).

I tried to use ImageJ to get the intensities of the in-focus partiles. But here are many problems I am facing:
1). How to pick up the in-focus particles? As the intensity cannot be used as a criterion, I use the particle size to select particles in focus. But I still have to use the threshold and this process will exclude some dim in focus particles. Is there any statistical method to select the in-focus particles?
2). I use the Analyze > Analyze particles to get the particle (mean) intensity. I am not quite sure how ImageJ calculate it (average of the particle occupied area?). But it will be very different values with different outlines are picked for particles.  I suppose that for the real intensity of a particle, it is better to put a Gaussian fit and calculate the integral area?  
3). As to RGB images, I want to calculate the intensity in each channel (R, G and B) for the same particle. But the images have to change to 8 bit grayscale. But what if I have a 12 bit image, and this calculation will lose the data I have in original image. This happens even for a 12 bit grayscale image. How can I get the real intensity (0- 2^12) without losing resolution?

Sorry for the long description but want to make it clear. Your help will be greatly appreciated.

Cheers,
Hui