Specularity detection method based on MSER

摘要

A new algorithm for specularity detection using a single image, which is based on the maximally stable extremal regions (MSER) detector, is presented. This algorithm takes full advantage of the characteristics of specularity and requires no assumptions. For this reason, it separates specularities very well regardless of the illumination condition and the surface material. To illustrate the effectiveness of the prposed algorithm, experiments are performed using the images of both dielectric and metal surfaces. Experimental results show that the presented approach outperforms the traditional methods from the point of view of sensitivity to correctness.