Preliminary study on vision-based pen-and-ink drawing by a robotic manipulator

摘要

This paper presents a robotic system that can perform automated pen-and-ink drawing based on visual feedback, using a proposed algorithm for stroke trajectory planning. The algorithm first converts the outlines of an input image to stroke trajectory according to the structural importance. Then, iterative hatching is carried out to convey both the tone and textures of the original image; in this process, visual feedback is employed to determine stroke positions, and local gradient interpolation is applied to guide stroke orientations. Finally, the drawing automatically terminates at the minimum point of a proposed criterion function, so that the drawing performance is not only robust to various input image tones, but also convenient to be tuned for a different style. Experimental results demonstrate that the proposed algorithm can create desirable pen-and-ink works in a robust way.