Correction of humidity effect for detection of human body odor

Abstract

Humidity strongly effects the sensitivity of odor sensors. It is therefore a major problem in the use of electronic nose (E-nose) for most applications including detecting human body odor from armpits, where humidity at the armpits can varies to a large extent due to various human activities. In this paper, we propose both hardware and software approaches to correct the humidity effect. The E-nose was designed to efficiently measure volatile organic compounds generated from human body and was most optimized if both the hardware and software corrections were employed. Principle component analysis (PCA) method was used for pattern recognition and discrimination of human body odor. After humidity correction, our special designed E-nose not only shows the capability in detecting human body odor, but it is also able to classify two different persons who have the same life style and activities.