A New Technique for Improving the Estimation of a Reflective Optical Color Sensor

Abstract

This paper proposes and experimentally demonstrates a novel technique of a precise color estimation by using a low cost reflective optical color sensor. The digital output signal of this sensor is a pulses series with a frequency proportional to the intensity of the incident light. In this sensor, three independent detector output signals with optimized spectral responsiveness allow discrimination between red, green and blue (RGB) light spectral components. The analytical expressions for RGB absorption efficiencies, which detect the color information, have been derived. An experimental set-up has been designed to determine the color difference, involving design of a portable color analyzer with wired and wireless computer interface. The detailed description and implementation of each design element are presented, while a developed algorithm is also discussed. The proposed algorithm apply a linear transformation to realize standardized colorimetric RGB and CIE31-XYZ color space coordinates responses. To that end, a correlation of the sensor outputs is established, a wide band correction matrix in old estimations is replaced by many narrowband correction matrices and the CIE standard curves of wavelengths are extracted and used to select a suitable narrowband correction matrix. Also, the sequential algorithm has been parallelized to reduce the response time in multiple measurement applications which need the big data processing. The experimental results in the vision spectrum confirm the theoretical analysis. Based on the findings of this experiment, the proposed system constitutes a high resolution and low cost solution for various color-sensing applications and colorimetric fields.