Binary local descriptors based on robust hashing

A robust hash, or content-based fingerprint, is a succinct representation of the perceptually most relevant parts of a multimedia object. A key requirement of fingerprinting is that elements with perceptually similar content should map to the same fingerprint, even if their bit-level representations are different. In this work we focus on the construction of discriminative binary local descriptors exploiting a combination of content-based fingerprinting techniques and computationally efficient filters (box filters, Haar-like features, etc.) applied to image patches. In particular, we define a possibly large set of filters and iteratively select the most discriminative ones resorting to boosting techniques. The output values of the filtering process are quantized to one bit, leading to a very compact binary descriptor. Preliminary results show that such descriptor leads to compelling results, outperforming SIFT in terms of truepositive / false-positive rate when using as few as 64 bits.