A study on the perceived quality of images displayed with color changes is presented. Initially, two hyperspectral images colors were represented in the CIE 1976 (L*a*b*) color space under the D65 standard illuminant. The colors of each image were divided into four clusters with the application of the K-Means algorithm. A new set of images was created changing only one of the four color clusters. The color change results from the application of a predefined chromatic error to the chromatic coordinates (a*, b*) of all pixels in the cluster. Errors of 6, 9, 12 and 15 ΔE* ab units were applied, using two different directions for each color cluster. These images were displayed for individuals visualization, that were asked to rank their quality based on their naturalness. The Mean Opinion Scores was computed and allowed to test and quantify the sensitivity to specific color changes.