Quantitative evaluation of performance through bootstrapping: edge detection

A new quantitative performance evaluation technique for computer vision systems is proposed. In real situations the complexity of input data and/or computational procedure can make the traditional error propagation methods infeasible. Using bootstrapping, a numerical technique for deriving statistical characteristics from a single sample, the authors perturb the nuisance properties of the input image to obtain distributions for the output variables. The performance thus is evaluated for the given input and system and not under simplifying assumptions. The task of edge detection is used as example.

[1]  John F. Canny,et al.  A Computational Approach to Edge Detection , 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  K. Mardia Statistics of Directional Data , 1972 .

[3]  L. J. Kitchen,et al.  The effect of spatial discretization on the magnitude and direction response of simple differential edge operators on a step edge , 1987, Comput. Vis. Graph. Image Process..

[4]  R. Tibshirani,et al.  An introduction to the bootstrap , 1993 .

[5]  Isaac Weiss,et al.  Smoothed differentiation filters for images , 1990, [1990] Proceedings. 10th International Conference on Pattern Recognition.

[6]  K. Ramesh Babu,et al.  Linear Feature Extraction and Description , 1979, IJCAI.

[7]  V. Ramesh,et al.  Automatic selection of tuning parameters for feature extraction sequences , 1994, 1994 Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.

[8]  Vinciane Lacroix,et al.  A Three-Module Strategy for Edge Detection , 1988, IEEE Trans. Pattern Anal. Mach. Intell..

[9]  Robert M. Haralick,et al.  Performance Characterization in Computer Vision , 1993, BMVC.