Generalized gradient: basic principles and example application

Abstract

This paper presents a generalized approach for computing image gradient. It is predominantly aimed at detecting unclear and in certain circumstances even completely invisible borders in large 2D and 3D texture images. The method exploits the conventional approach of sliding window. Once two pixel/voxel sets are subsampled from orthogonal window halves, they are compared by a suitable technique (e.g., statistical t-test, SVM classifier, comparison of parameters of two distributions) and the resultant measure of difference (e.g., t-value, the classification accuracy, skewness difference of two distributions etc.) is treated as the gradient magnitude. The bootstrap procedure is employed for increasing the accuracy of difference assessment of two pixel/voxel sets.