生活支援ロボットのためのしわ特徴に基づく衣類発見法 ○ 山崎公俊 稲葉雅幸(東京大学 ) A Cloth Detection Method Based on Wrinkle Features

This paper describes an image based clothes detection method. The method enables daily assistive robots to find clothes placed on real environments. No prior knowledge related to shape model or color information is needed because wrinkles on a cloth are extracted as image feature. The feature is constructed through SVM learning by using the result of Gabor filtering. Clothes extraction based on the feature is combined with graphcut, image regions are divided into cloth regions and background. Effectiveness of the proposed method was illustrated through tidying clothes up by a real robot in living room.

[1]  Hirochika Inoue,et al.  Rope handling by a robot with visual feedback , 1987, Adv. Robotics.

[2]  Rajiv Dubey,et al.  Robotic Fabric Handling for Automating Garment Manufacturing , 1983 .

[3]  Vladimir Kolmogorov,et al.  An experimental comparison of min-cut/max- flow algorithms for energy minimization in vision , 2001, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[4]  A deformable model driven method for handling clothes , 2004, Proceedings of the 17th International Conference on Pattern Recognition, 2004. ICPR 2004..

[5]  Ehud Kroll,et al.  Toward assembly sequence planning with flexible parts , 1996, Proceedings of IEEE International Conference on Robotics and Automation.

[6]  Hitoshi Akami,et al.  Robot Hand with a Sensor for Cloth Handling , 1989 .

[7]  Andrew P. Witkin,et al.  Large steps in cloth simulation , 1998, SIGGRAPH.

[8]  Matthijs C. Dorst.  Distinctive Image Features from Scale-Invariant Keypoints , 2011 .

[9]  Vladimir Vapnik,et al.  Statistical learning theory , 1998 .