Regression Method Based on ACE for Humidity Measurement on Colorimetric Sensor Array

Abstract

Colorimetric sensor is designed for indicating category or quantity of some substance via colour changing. It is essential to establish a mapping that can convert colour information into measurement. In this paper a regression method was proposed for relatively humidity measurement on a type of phage-based colorimetric sensor array with multiple array units. The method was based on alternating conditional expectations (ACE) algorithm, dealing with extracting features from highdimension colour information. And support vector regression (SVR) was applied to achieve humidity measurement. Compared with extracting features by PCA, the proposed method has better accuracy of recognizing relative humidity beyond 0.85 when dimensions of colour information increase. It will enable colorimetric sensor array for quantitative measurement under high-dimension colour information.