Introducing colors in barcodes and QR codes would significantly extend the data capacity of such markers, but still today represents a major challenge because of the large variety of color distortions often occurring in real world scenes. To address this problem, several constraints on the light and on the marker structure and colors have been introduced in the past, limiting the applicability range of the color markers. To overcome this issue, here we propose (i) a novel model for color markers with the only constraints of being composed of matte material and not differing from each other in color intensity solely and (ii) a HDR imaging system for the illuminant invariant recognition of such markers. In our system, an optical sensor acquires marker colors over a linear dynamic range of about 105 dB, while a microcontroller implements a nearest-neighbor classification in an illuminant invariant color space. We tested a prototype of this system on a set of 15 color markers printed on matte, white paper and acquired under 18 lights, reaching on average 99.14% of correct classification.