○呉 榮軒(東工大),齋藤 豪(お茶の水女子大/東工大)
A shadow appears while an object obstructs the light rays from light source. Therefore, a shadow defines powerful constraints between the light source, object, and surface on which it is cast. We present a novel objective function to estimate the 3D model of objects, which aims to create a 3D model casting the same shadow as seen in input image. Our method requires single input image and several user annotations, and is applicable to many different objects as long as the object casts hard shadow in input image. We tested on photographs of indoor and outdoor scenes illuminated by various types of light source and visually plausible shadows are synthesized with the estimated 3D models.

footer 情報処理学会 セキュリティ プライバシーポリシー 倫理綱領 著作権について