Table 3.

Results for the linear mixed model analysis over images.

	Self-similarity		Complexity		Anisotropy		Birkhoff-like measure		HSV color hue		HSV color saturation		HSV color value		Aspect ratio
	RC	p	RC	p	RC	p	RC	p	RC	p	RC	p	RC	p	RC	p
Cluster 1	−0.345	0.308	−0.002	0.790	−154	0.378	−0.455	0.010	−0.664	0.029	1.329	<0.001	0.677	0.049	0.247	0.124
Cluster 2	−1.170	0.001	0.013	0.120	−282	0.107	−0.555	0.002	−1.425	<0.001	0.245	0.549	0.932	0.007	0.157	0.317
Cluster 3	0.022	0.946	−0.005	0.501	−319	0.069	−0.665	<0.001	−1.106	<0.001	0.827	0.045	0.214	0.529	0.200	0.208
Cluster 4	−1.603	0.000	0.007	0.401	−454	0.010	−0.291	0.086	−1.928	<0.001	0.876	0.033	0.691	0.044	0.655	<0.001
Cluster 5	−1.023	0.003	0.019	0.023	−436	0.014	−1.185	<0.001	−0.571	0.059	−0.243	0.552	0.783	0.023	0.301	0.065
Cluster 6	0.512	0.133	0.001	0.941	−299	0.089	−0.575	0.002	−1.456	0.000	0.729	0.076	0.345	0.312	0.491	0.005

Displayed are the regression coefficients for the difference of the effect of the statistical image properties on subjective ratings between each cluster and the reference cluster (cluster 7). The regression coefficients reflect the interaction between statistical image properties and cluster membership. RC, regression coefficient; p, p value.