Given the image is stimulate and have a uniform noise, I should be able to calculate the SNR by simply dividing the mean signal by SD of noise. This is what I did, in more details;
Each image was masked into two regions, a rectangular in the middle contained the peaks and another mask excluding the first ROI. The SNR was determined by first, subtracting the noise-mean (selected off the signal area in the middle of the image) by the mean of 30 peaks local-maxima-amplitude (one particle per spot). This value was then divided by the standard deviation of noise to calculate the SNR. Note that all the zeros were excluded from this analysis.