Content area
Abstract
As a striking art form, an image mosaic uses a collection of small images to construct a larger image that suggests a whole image seen at a distance, but the individual images are visible when observed closely. In this thesis, we introduce the general procedure for creating an image mosaic, elaborate several distance algorithms based on the color metrics to create image mosaics, and then propose a novel method based on the color coherence vectors of images that contain more spatial information compared with other color distance methods. Experimental results show the mosaic effects.
Another contribution of this thesis is that we extend the general image mosaic by using image shift, image rotation, etc., transform techniques and non-uniform tile image overlap, as well as present and analyze image mosaic optimization methods. Experimental results exhibit some distinct aesthetic mosaic effects.