A method, system and apparatus for determining and modifying saliency of a visual medium are provided. The method, system and apparatus may obtain saliency values for a visual medium based on a plurality of visual channels. The saliency values may be obtained based on at least one of computer-generated modeling, user-specified input and eye-tracking. The method, system and apparatus may aggregate the obtained saliency values and classify regions of the visual medium based on the aggregated saliency values. The visual channels may include one or more of absolute mean curvature, a gradient of mean curvature, a gradient of color intensity, color luminance, color opponency, color saturation, lighting and focus. When calculating mean curvature, the method, system and apparatus may calculate a change in mean curvature for a plurality of vertices around a region and displace the vertices in accordance with the calculated change in mean curvature to change a saliency of the region.