Ridge Extraction Using Adaptive Direction Information and Tracking Techniques

The ridge is a type of edge. However, unlike most types of edges, the ridge acts like a Laplacian-of-Gaussian (LoG) function as a step function. However, since using the LoG is similar to the second-order differentiation, the result is largely affected by noise, and the detail is over-stressed. To solve the problem, a robust ridge extraction algorithm was developed in this study. First, the adaptive direction information was generated and refined by an iterative extension algorithm. Moreover, with accurate information, a tracking method connected the broken ridge line. Experiments showed that, with the proposed algorithm, the ridge was extracted even if the image was highly interfered with by noise. Regardless that the ridge lines were too close or crossed with one another, the ridge was extracted well.