Computer Vision, Graphics, and Image Processing
Dynamic threshold determination by local and global edge evaluation
Graphical Models and Image Processing
Robust Visual Method for Assessing the Relative Performance of Edge-Detection Algorithms
IEEE Transactions on Pattern Analysis and Machine Intelligence
Edge Detection and Ridge Detection with Automatic Scale Selection
International Journal of Computer Vision
Edge detector evaluation using empirical ROC curves
Computer Vision and Image Understanding - Special issue on empirical evaluation of computer vision algorithms
Comparison of edge detector performance through use in an object recognition task
Computer Vision and Image Understanding - Special issue on empirical evaluation of computer vision algorithms
Regularized Laplacian Zero Crossings as Optimal Edge Integrators
International Journal of Computer Vision
Device Space Design for Efficient Scale-Space Edge Detection
ICCS '02 Proceedings of the International Conference on Computational Science-Part I
Distortion Analysis on Discrete Laplacian Operators by Introducing Random Images
ICIG '04 Proceedings of the Third International Conference on Image and Graphics
IEEE Transactions on Image Processing
Quantitative error measures for edge detection
Pattern Recognition
Hi-index | 0.00 |
A new evaluation technique is presented to enable edge sensitivity analysis with respect to angular orientation and displacement errors for edges located by discrete zero-crossing operators. The technique is validated by using a Gaussian edge model and is shown to provide an effective mechanism for characterising the quality of second derivative feature detection operators in terms of quantitative measures of correctness of edge location and orientation. The technique applies a finite element interpolation to the output values of the discrete operator in order to extract sub-pixel level information about zero-crossings; in general, the displacement and orientation of a local line segment along which the line integral of the output interpolant is zero may then be readily found as the solution of a pair of simultaneous algebraic equations. A significant advantage over earlier edge sensitivity techniques is that the method does not require the use of a supplementary first derivative operator for gradient approximation. The method can therefore be used to make direct comparisons between zero-crossing operators in terms of basic performance standards without reference to particular test images; such standards are also important as they form the necessary basis for investigating the potential for the use of proxies for operator performance in relation to subsequent higher-level image processing tasks.