Photo matching using skin color histogram without and with Instagram-like modifications

Social networking has become a part of our lives. Web platforms enable communication and exchange of different media for private and business purposes, where beauty or Instagram-like filters are applied for better representation. Among face characteristics, skin color is often used for recognition or matching. In this paper feature vectors corresponding to unmodified and modified query images are based on two color space domains (HSV and Lab). Seven different distance metrics are applied for matching purposes, as well several Instagram related filters, like sepia and kelvin. The results show that image filtering affect color skin based approaches in information retrieval.