Robust Curve Detection using a Radon Transform in Orientation Space Applied to Fracture Detection in Borehole Images

We present a novel approach to parameterised curve detection. The method is based on the generalised Radon transform, which is traditionally applied to a 2D edge/line map. The novelty of our method is the mapping of the original 2D image to a 3D orientation space, which then forms the input for the Radon transform. The orientation space representation offers important advantages. It can represent multiple intersecting structures, contains local orientation information, and offers a better SNR. A Radon transform applied to this space is automatically constrained by local orientation information, thus reducing the amount of false detections. The method is applied to fracture detection in borehole images. The challenge is to detect intersecting (partial) sinusoidal curves in a heterogeneous and noisy background. The orientation space representation has proven to be instrumental for good detection of the fractures.