Employing Fusion of Learned and Handcrafted Features for Unconstrained Ear Recognition

摘要

We present an unconstrained ear recognition framework that outperformsstate-of-the-art systems in different publicly available image databases. Tothis end, we developed CNN-based solutions for ear normalization anddescription, we used well-known handcrafted descriptors, and we fused learnedand handcrafted features to improve recognition. We designed a two-stagelandmark detector that successfully worked under untrained scenarios. We usedthe results generated to perform a geometric image normalization that boostedthe performance of all evaluated descriptors. Our CNN descriptor outperformedother CNN-based works in the literature, specially in more difficult scenarios.The fusion of learned and handcrafted matchers appears to be complementary asit achieved the best performance in all experiments. The obtained resultsoutperformed all other reported results for the UERC challenge, which containsthe most difficult database nowadays.