Abstract
In this paper, we propose a flexible model for edge representation. The model is able to capture important characteristics of edge profiles in digital imagery such as the blur level of the edges and the amounts of gutters and humps around the edges. First, the model for edge profiles is developed based on the Dolph-Chebyshev function and the procedure of estimating the edge model parameters under a least - square framework is presented. Then, the validity of the proposed model is proved experimentally using edge profiles extracted from images with various blur and noise levels. The model is compared with the single and double Gaussian-based models and it is observed that the proposed model consistently produces the lowest root-mean-square-error (RMSE) on all tested edge profiles. Finally, the procedure for edge detection in two-dimensional images using the proposed model is derived and it is shown that our edge detector performs better than the single and double Gaussian-based edge detectors in terms of F-measure.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call