Human positioning and tracking in smart environments using colour pattern matching

摘要

The gold standard for detection and tracking of a human target in an indoor setting is Background Subtraction (BS). However BS has numerous problems when utilised indoors. Colour pattern matching may offer a viable and superior alternative to BS whilst enhancing privacy and person-device independence. Results obtained from this study demonstrate that, when using colour pattern matching to target a person's clothing, detection rates of 97.28% can be maintained across the electromagnetic spectrum of visible light, and during instances of acute differentiation in ambient illumination whilst allowing for a privacy-enabling blurring technique to be applied to all frames.