Measuring ASTM grain size by automatic image analysis is often made difficult by the inability of the image analyzer to detect grain boundaries accurately. For this reason, a logic-based method for grain boundary reconstruction has been developed. ASTM G was measured at various grain sizes both before and after image processing. Automatic image analysis was used to measure the grain size on actual microstructures as well as on standard grain size charts that had grain boundaries intentionally removed. ASTM G was measured by both planimetric and intercept methods. The grain size calculated on the basis of area was much more sensitive to incomplete boundaries than was the calculation based on intercepts. Nevertheless, the software was able to restore the boundaries, and the correct grain size was calculated.

In addition to grain boundary reconstruction, software has been developed that uses artificial intelligence to detect twin boundaries so that they can be removed before grain size measurement. The software identifies twins at several levels of confidence for the user to select only those structures that are actually twins.