Image enhancement based on Log-Gabor filter for noisy fingerprint image

Abstract

Image enhancement is a technique to improve the quality metrics of a distorted image. In forensic laboratory, if the input is noisy fingerprint image then image enhancement plays the crucial role to authenticate the verification stage. In this work, the author proposed to construct a filter based on Log-Gabor theory for enhancement of high noisy (Gaussian Noise) fingerprint image. To construct the filter, first initialize all filter parameters and to overcome the wrap around effect, first calculate the sine and cosine differences along with angular distance. When the filter is constructed, do the convolution with FFT of input noisy image. The resultant enhanced image attained PSNR and MAE (Mean Absolute Error) 7.9230, 101.728 dB, respectively. The comparison values of quality metrics for Mat lab inbuilt function (wiener and median filter) are 7.3291, 7.3281, and 101.977, 101.986 dB respectively for same test image. In this work, there is a little improvement in PSNR value which play significant role in verification stage.