Transform domain edge enhancement of digital radiographs

Abstract

An enhanced digital radiograph is obtained by multiplication between the digital radiograph and the digital Fourier transform of the second derivative of the Gaussian function ( Delta /sup 2/g) and inverse transformation, and the subsequent histogram equalization. Because image data includes only real values, the purely real-numbered transform, called the fast Hartley transform (FHT) is used to perform a digital Fourier transform of the image data. Preliminary results suggest that the transform of the Laplacian of the Gaussian may provide a more effective edge enhancement for medical imagery than space-domain convolution techniques such as the Sobel operator.<>