Automatic fingerprint recognition is an on-demand system in most of the authentication devices. The security system only has limited memory space due to the expensive nature of memory elements. However, as more persons are included in the repository, the size of the database has grown extensively. So the memory storage is a challenging problem while storing high quality or too many images. This paper proposes a Pattern Optimization from Subset Tree (POST) image coder for fingerprint image compression which improves the coding efficiency. Initially, extract the ridge based features after the median filter smoothing for the fingerprint identification system. The features thus formed as a probability map and further processed for compression. POST identifies the significant coefficients from the subdivided tree coefficients and the resultant stream of bit patterns are optimized for low bit rate. Finally, the experimental results demonstrate the performance measures in terms PSNR, CR, bpp, and EER for the compression scheme.