Ear Detection Based on Arc-Masking Extraction and AdaBoost Polling Verification

This paper proposes a simple but practical 2D ear detection algorithm based on arc-masking candidate extraction and AdaBoost polling verification. In the first half phase of the proposed ear detection algorithm, a few ear candidates are extracted by arc-masking edge search followed by multilayer mosaic and orthogonal projection histogram. Then, in the second half phase, the most likely ear candidate is picked out by rough AdaBoost polling verification. Experimental results show that the proposed ear detection algorithm can achieve a bit higher detection hit rate and much lower detection false alarm rate than conventional AdaBoost ear detection algorithm with Haar-like features under various pose rotation conditions.