
There are a number of pricy softwares (if you haven't written your own software) that does it (I used ImagePro, a $5,000 for a single license). But the problem is, the sizing data relies heavily on how the image was taken. For this very reason I advice you to read the recent ASAP paper on "particle size determination using TEM: A Discussion of Image Acquisition and Analysis for the Novice Microscopist". Pryz et al. (Buttrey group in U Delaware) gives a good ride on how inaccurate a TEM picture would become. I feel like this paper may become a landmark.
Thanks to Dr. Aasheesh Srivastava for the tip.