Depiction and Perception of Shape in Line Drawings

点击翻译按钮获取中文摘要

Depiction and Perception of Shape in Line Drawings Doug DeCarlo, Manish Singh and Matthew Stone Rutgers UniversityAbstractBecause they can focus on what is essential about a scene, line drawings are a fundamental way of conveying shape that are often preferred over realistic representations,. This research couples the development of computer-based methods to synthesize line drawings with experiments on human observers that reveal how line drawings are perceived and interpreted. The ultimate goal is to produce depictions of 3D shape that are easier to understand. By improving line drawings, this research improves a source of communicative imagery that could become ubiquitous---appearing in better maps, more flexible and customizable technical and medical illustration, more broadly accessible scientific visualization, and many other applications. At the same time, new interactive techniques promise to make computer representation of 3D objects accessible to a wider range of users for a wider range of tasks and on a wider range of platforms.This research involves three interrelated thrusts: (1) Computational studies are used to characterize the ambiguity inherent in interpreting line drawings and to develop new line styles that aim to reduce this ambiguity; (2) Experimental studies that probe the visual perception of 3D shape are used to discover what shape information human viewers gather from line drawings; and (3) The investigators are developing applications that exploit a perceptual model derived from the experimental studies in order to guide the placement of lines in drawings, and to guide the interpretation of hand-drawn sketches for interactive model construction.