ALOHA: An efficient binary descriptor based on Haar features

Abstract

This paper introduces ALOHA (Aggregated LOcal HAar), a compact and efficient binary descriptor based on a small number of intensity difference tests to represent an image patch as a binary string. ALOHA uses a set of features, reminiscent of Haar basis functions, to group pixels within a patch centered on a keypoint. It has been compared with two version of BRIEF, the current best in class short binary descriptors. Even though the matching time for both descriptors is identical, ALOHA is faster to compute, and ensures better matching accuracy and discrimination capacity. © 2012 IEEE.