A Novel Approach for Low‐Quality Fingerprint Image Enhancement Using Spatial and Frequency Domain Filtering Techniques

Fingerprints find use in many applications such as law enforcement, health care, and education. Generally, the images of the finger prints are not good quality images because of different input contexts. Most of the existing techniques for enhancing fingerprint images use any one approach: frequency or spatial domain. These techniques are not able to ultimately enhance the complex ridge structure of the images. This chapter proposed a novel algorithm for enhancing the quality of the fingerprint image using frequency and spatial domain both. The frequency and orientation estimations were fed to the Gabor filters bank in the spatial domain to recover the damaged regions and improve the image contrast. The output of the Gabor filter was fed to the frequency domain band-pass filter. The frequency and orientation estimates were done again using the spatial domain–filtered image. The band-pass filter uses these re-estimated estimates to enhance the image further significantly. The results of the proposed algorithm were compared with some of the state-of-the-art techniques over the FVC2004 database. The texture descriptors and minutiae ratios were used to compare the results. Proposed algorithm gives the best texture and achieves a true minutiae ratio (TMR) of 94.82% with a thinning feature extraction technique and a TMR of 90.45% with the mindset feature extraction technique, and low other minutiae ratios as compared to the other state-of-the-art techniques. Moreover, it gives the best results for all the sub-databases of the FVC2004 in comparison with the existing techniques.