A new methodology for evaluation of edge detectors

Abstract

This paper defines a new methodology for evaluating edge detectors through measurements on edginess maps instead of on binary edge maps as previous methodologies do. These measurements avoid possible bias introduced by the applicationdependent process of generating binary edge maps from edginess maps. The features of completeness, discriminability, precision and robustness, which a general-purpose edge detector must comply with, are introduced. The R, DS, P and F AR-measurements in addition to P SNR applied to the edginess maps are defined to assess the performance of edge detection. The R, DS, P and F AR-measurements can be seen as generalizations of previously proposed measurements on binary edge maps. Well-known and state-of-the-art edge detectors have been compared by means of the new proposed metrics. Results show that it is difficult for an edge detector to comply with all the proposed features.

Gradient-based enhancement of tubular structures in medical images

“…This property is related to discriminability as described in Moreno et al (2009). This property can be assessed by comparing the mean vesselness inside a generalized cylinder of radius δ 1 to the one at the region between that cylinder and a larger concentric generalized cylinder (cf.…”

Tensor Voting for Robust Color Edge Detection

“…Many evaluation methodologies have been proposed to evaluate performance at the second e.g., [26,10,4,15] and third points [29,2]. On the other hand, to the best of our knowledge, the methodology introduced in [22] has been the first attempt to measure performance at the first point.…”

“…Recently, we have also introduced a general methodology to evaluate edge detectors directly in gray-scale [22]. This methodology avoids possible biases generated by post-processing of edginess maps by directly comparing the algorithms in gray-scale.…”

Determination of Additional Aperture in Non-Metal Sewer Pipes by Image Processing

“…So, filters output would be grater (either negative or positive) when there are outstanding variations in the values of pixel in the right and left sides of the pixel site. A set of weights, which includes some smoothing in averaging over 3 rows is shown by [11,12]:…”