Demosaic-free color detection for particle images

Abstract

We propose a robust method to detect colors, positions, and sizes of particles on color particle images. The method is free of color artifacts originating from the demosaic process of Bayer raw images. We test the method using synthetic color particle images to quantitatively evaluate its performance and demonstrate its capacity for precise extraction of colors while refining particle positions at sub-sub-pixel level. We demonstrate the method’s applicability by applying it to different types of particle images from laboratory experiments of color particle tracking velocimetry and liquid crystal thermometry. The method standardizes color detection on particle images and promises the improvement in final performances of these color-utilizing measurements.