Abstract

Most image representations and device calibrations are based on the properties of the human visual system, described in CIE standards such as XYZ or CIELAB. But during color image transmission, an image may be input by devices whose wavelength selectivity differs significantly from the human eye. If image transmission procedures ignore these differences, information can be lost or distorted during device-to-device transmission. We are exploring image representations and calibration procedures based on the physical properties of reflectance and absorbance of light. We view device calibration and communication as a physical estimation problem in which receivers estimate and communicate about the spectral properties of the input. For initially linear devices, such as CCD devices or the initial encoding by the human eye, interdevice color communication is analogous to the problem confronting observers looking at a scene from different points of view. Each observer sees a different projection of the scene, but if the observers know enough about their environment they can infer what the other observers are seeing. So too, each color device may sense the spectrum slightly differently, but if we have enough knowledge of the likely functions of wavelength interdevice communication can succeed.

© 1991 Optical Society of America