Naturalness is one of important image appearance attributes to achieve a high-quality image, but few attempts have been made to quantify it. The current study introduces modelling image naturalness based on visual phenomena that had a large impact on perceived image naturalness. Psychophysical experiments were conducted under a dark surround. Twelve observers were asked to assess image naturalness using the categorical judgment method. Eight colour images were manipulated to produce variations by rendering lightness, chroma and sharpness. For modelling image naturalness, memory colours and four factors revealed to be important in the results of the psychophysical experiment were considered. A naturalness index determined from memory colours alone was not sufficient to predict perceived naturalness. The four key factors were however successfully modelled, based on parameters derived using the CIECAM02 colour appearance model and the CAM02-UCS uniform colour space for image naturalness.