The purpose of the present invention is to provide a pattern measurement device which evaluates quantitatively and with high precision random patterns such as finger print patterns. In order to fulfill this purpose, a pattern measurement device which measures the pattern on a sample on the basis of an image acquired by a charged particle beam is proposed which selectively extracts linear or linearly approximable parts of the pattern on the sample, and outputs at least one of the following: the measurement of the distance between the extracted parts, the ratio of said extracted parts in a prescribed region, and the length of said extracted parts. Further, as a more specific embodiment, a pattern measurement device is proposed which calculates a frequency depending on a distance value between extracted parts, and outputs, as a pattern distance, distance values for which said frequency fulfills a prescribed condition.