Abstract In reservoirs with complex structure, such as turbidite sequences, rock fabrics exhibit great variations both vertically (example-Bouma sequences) and in lateral direction. In such formations, patterns observed on lithological logs can show considerable differences. Constructing rational and meaningful correlation models consistent with all the lithological signatures can be an enormous task before conceptualization of data on a 3-D geologic model. The proposed method can serve as an important pre-processing step to minimize the requirements of the human expert input in the sub-marker detection for substantial number of well logs. The automated techniques of pattern classification enhance the characterization and identification of stratigraphic features of laminated type reservoirs. The approach proposed in this paper works on basic lithological logs and marker information. It comprises noise filtering and pattern recognition that lead to identification of reservoir compartments. This helps in delineation of the lateral continuity and discontinuity of reservoir sand and shale laminations.