In this paper we have presented an algorithm allowing extracting automatically frequency components from color images. Through analysis of the image, image to sound frequency transformation was realized. The dominant layer of the color image was used in order to map the image frequencies to a pertinent musical scale from a psychoacoustic point of view: Major for joy, pentatonic minor for relaxation, hidjaz for an eastern style of music. The observations showed that further criteria and methods are needed to be developed in order to cover negative feelings like stress or anger. With the usage of pattern recognition in this process it can be possible to associate also musical patterns with recognized objects or scenes in order to augment the quality of the resulted sound from a musical point of view