Optimizing the FBI fingerprint compression algorithm for BLACKFIN processor

This paper presents an effective method to convert the conventional floating-point algorithm of the FBI Fingerprint Compression Into a fixed-point algorithm suitable for digital signal processors. Wavelet transform, the most computationally expensive component of the technique, is optimized to reduce the computational burden significantly. The conversion and optimization are tailored to fit the advanced structure of the Blackfln processor. The novel algorithm also retains an acceptable quality of the reconstructed image.