A regular grammar approach for the chance collection of partial fingerprint pattern classification is described in this paper. The procedure involves the division of a fingerprint pattern into quadrants and the transformation of each quadrant into sampling matrices indicating dominant direction of the ridges for each subpicture. These sampling matrices are used for feature extraction and a class of regular grammar is then used to describe each quadrant of the partial fingerprint pattern. The classification is then accomplished by means of finite automata for each quadrant.