Novel Statistical Approaches to the Quantitative Combination of Multiple Edge Detectors

This paper aims at describing a new framework which allows for the quantitative combination of different edge detectors based on the correspondence between the outcomes of a pre-selected set of operators. This is inspired from the problem that despite the enormous amount of literature on edge detection techniques, there is no single one that performs well in every possible image context. The so-called Kappa Statistics are employed in a novel fashion to enable a sound performance evaluation of the edge maps emerged from different parameter specifications. The proposed method is unique in the sense that the balance between the false detections (False Positives and False Negatives) is explicitly assessed in advanced and incorporated in the estimation of the optimum threshold. Results of this technique are demonstrated and compared to individual edge detection methods.