A Motion Comic Creation with Monochromatic Background Extraction from a Comic Image

摘要

We propose a method for creating a motion comic with monochromatic background extraction from a comic image. The targets of the background are white, monochromatic screen tone, and solid black, and are extracted with an image processing. The extracted background can be applied to an animation such as translation, expansion, and dissolving. The background extraction method consists of a saliency map method, speech balloon extraction, and posterization with a pixel value of an input image. In our experimental evaluation with four comics, an accuracy of correction extraction is 49.2 %. The animation was manually generated with the background extraction, and a possibility for an attractive content is confirmed.