Tensor Voting for Robust Color Edge Detection

Abstract

This chapter proposes two robust color edge detection methods based on tensor voting. The first method is a direct adaptation of the classical tensor voting to color images where tensors are initialized with either the gradient or the local color structure tensor. The second method is based on an extension of tensor voting in which the encoding and voting processes are specifically tailored to robust edge detection in color images. In this case, three tensors are used to encode local CIELAB color channels and edginess, while the voting process propagates both color and edginess by applying perception-based rules. Unlike the classical tensor voting, the second method considers the context in the voting process. Recall, discriminability, precision, false alarm rejection and robustness measurements with respect to three different ground-truths have been used to compare the proposed methods with the state-of-the-art. Experimental results show that the proposed methods are competitive, especially in robustness. Moreover, these experiments evidence the difficulty of proposing an edge detector with a perfect performance with respect to all features and fields of application.