This paper proposes a fingerprint image enhancement algorithm that can be mapped onto the compact pixel-parallel architecture, such as a fingerprint identification chip that senses and identifies fingerprints by itself. The algorithm is composed of two parts. First, ridges are enhanced by extracting their center lines and removing white noises using the center line image. Then valleys are enhanced by detecting areas where they are thin and disconnected and dilating them at these areas. Both enhancements are done with the structures of the fingerprint maintained. The experimental results show fingerprint identification using the proposed image enhancement algorithm becomes much better, that the error ratio becomes a quarter of that of the conventional method, and the execution time is 0.14 msec when the proposed algorithm is performed on the array of processing elements in the fingerprint identification chip.