Classifying many-class high-dimensional fingerprint datasets using random forest of oblique decision trees

Abstract

Classifying fingerprint images may require an important features extraction step. The scale-invariant fea-ture transform which extracts local descriptors from images is robust to image scale, rotation and also to changes in illu-mination, noise, etc. It allows to represent an image in term of the comfortable bag-of-visual-words. This representation leads to a very large number of dimensions. In this case, ran-dom forest of oblique decision trees is very efficient for a small number of classes. However, in fingerprint classifica-tion, there are as many classes as individuals. A multi-class version of random forest of oblique decision trees is thus pro-posed. The numerical tests on seven real datasets (up to 5,000 dimensions and 389 classes) show that our proposal has very high accuracy and outperforms state-of-the-art algorithms.