Enhancing Oriented Pattern Using Adaptive Directional FFT Filter

It is well known that the output of Fourier Transform in general is orientation dependent. As such, by applying a two-dimensional (2D) discrete Fast Fourier Transform (FFT) onto an image with oriented pattern, the output image will show a clustering around a specific orientation. By strengthening the points located in the cluster and suppressing the points outside of this cluster, the oriented pattern in the image can be enhanced. This paper describes a new method in which a bandpass filter is coupled with an adaptive ‘angular filter’ to achieve directional filtering. This filter is then applied onto the twodimensional FFT output to enhance the image. The bandwidth of the angular filter is changed adaptively in proportion to the confidence level of the computed orientation of the desired pattern. Such adaptive changes in the angular bandwidth prevent excessive filtering where consistency of the orientation is low. The proposed approach is used to enhance fingerprint images. Experimental results show the desirable features of the proposed approach.