Hi,
often this kind of problem can be solved with preprocessing the image; for example, is filtering the image before edge detection helpful? If these dots come from outside of the domain of interest, maybe these regions are characterized by different properties you can use to eliminate these edges from the image before running edge detection. Please post an image before edge detection too, so the challenge can be clearer.
Also, edge with "canny" method comes with options, maybe altering them can prove useful.
Do you need all the small edges within the domain of interest, or would merging them be helpful / or at least won't interfere?
