The generalized Orientation Field Transform

We extend earlier work on the Orientation Field Transform as an effective method for enhancing curve like features in images. Whereas the original Orientation Field Transform was formulated for straight lines, we generalize the formalism to general curves, resulting in improved accuracy for detecting circular objects. The method only requires one parameter, the curve width, to be selected by the user and does not require any preor post processing, even for noisy data sets with low contrast. We demonstrate the method for several electron microscopy tomograms.