A measurement tool connects to an automatic inspection machine for identifying and measuring microscopic dimensions such as area, diameter, height and line width of defects and lines of a photographic mask. An operator draws a rough region of interest around a feature and the tool automatically identifies the feature and calculates its dimensions. For features less than one micron in size, the size of light photons interferes with measurement, so a non-linear polynomial calibration curve is developed for each machine. Features of known sizes are measured on a production machine to produce a calibration curve for each type of defect or line. Features of unknown sizes are measured on the same machine and the measured size in pixels are calibrated using the calibration curve to return a more accurate reading in microns. To determine a dimension, the type of a feature is determined by using a bounding box and light transitions of an intensity profile of the feature; multiple regions of interest are developed for each feature to accommodate angled lines; columns of pixels in each region of interest are summed to produce a light intensity distribution profile for each region of interest; total flux is determined from each profile and the best flux measurement is used to calculate a dimension of the feature. A good image of a feature is obtained by subtracting a reference image from an original image if a profile is of low quality. For lines separated by less than one resolution unit, a full-width half-maximum technique is used to calculate line width.