Reconnaissance tasks and the detection of camouflaged targets can be improved if the different surface reflection conditions of natural and artificial surfaces can be discriminated in multispectral images. Analogue to the visible, colorimetry in the shortwave infrared spectral range has been presented in praxis and theory in . In this presentation we will discuss the influence of a four filter set onto the capability of the system to discriminate different spectral characteristics of two objects. An optimal choice of filters is presented in the sense of a homogenous discrimination characteristics across the SWIR band, taking into account the SWIR solar spectrum. Based on this choice two development issues were investigated: the accuracy of the color values with respect to measurement noise and the display of SWIR color images. As a figure of merit for color value accuracy the Noise Equivalent Wavelength Difference is introduced describing the minimum color difference that can be resolved/discriminated from the noise floor. Due to the lack of a physiologic counterpart as known from the visual colorimetry where the eye is used as reference for the three color channels, we will showcase a model for transforming and displaying SWIR colors in the R-G-B color space.