Fingerprint Image De-noising through Decimated and Un-decimated Wavelet Transforms (WT)

This recent technologies for recognizing fingerprints have proven as regards as reliable for security purposes. But, the efficient recognition is depending on the quality of fingerprint image and it is a complex computer problem while dealing with noisy and low quality images. The most common methods used to acquire the fingerprint images do not need expertise, but highly distorted images are still possible because of dryness of skin, skin disease, dirt or humidity. Therefore such type of images must be de-noised for recognition. Many researchers have proved advantages of discrete wavelet transform (DWT) for image de-nosing. DWT in image de-noising has limitations due to its non-invariance in time/space. The time invariant property of SWT is useful for de-noising the image. In this paper we have studied effectiveness of orthogonal and bi-orthogonal wavelet transforms with different decomposition level using SWT. And the quality of de-noised image is evaluated through structural distortion measurement and also measured the computational processing time at each level. Our results show coifl at level 5 gives the best results.