Using the generalized Radon transform for detection of curves in noisy images

Abstract

In this paper the discrete generalized Radon transform will be investigated as a tool for detection of curves in noisy digital images. The discrete generalized Radon transform maps an image into a parameter domain, where curves following a specific parameterized curve form will correspond to a peak in the parameter domain. A major advantage of the generalized Radon transform is that the curves are allowed to intersect. This enables a thresholding algorithm in the parameter domain for simultaneous detection of curve parameters. A threshold level based on the noise level in the image is derived. A numerical example is presented to illustrate the theory.