Figure-ground image generation using contour matching and rigid shape deformation
SIGGRAPH Asia 2013 Posters(2014)
摘要
Figure-ground perception is the perceptual experience by which human observers assign a contour separating two regions to one of regions. The region owning the contour is identified as figure which has definite shape and appears near to the observer while the other region, the ground, appears shapeless near the contour and seemly occluded by figure. Determine which portions of the image correspond to figure, also called figure assignment, which is a key procedure in the perceptual organization. Some related works include [Tong et al. 2011; Yoon et al. 2008] have been proposed computer methods to help generate art images. However these papers concentrate on nested image or hiding objects, while our work is very different. We are inspired by Rubins vase illusion, many artists are devoted to create similar binary images or silhouette art with strong ambiguous figure-ground perception. We propose a semi-automatic system that composes an ambiguous figure-ground images from two binary images with one served as figure and the other as ground, our system allows artists significant flexibility for their creativity.
更多查看译文
关键词
animation
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络