HCC: Small: Automatic Simplification Methods for Tactile Graphics

点击翻译按钮获取中文摘要

Unfortunately for individuals who are blind and visually impaired, although access to text has greatly improved with computers, the computer has also greatly facilitated the ease of producing, storing, transmitting and using graphical representations to convey information. This has created a serious obstacle for the millions of people in the United States who are visually impaired or blind, because graphics are now ubiquitous and used in many instances as the sole information presentation method. Although presenting graphical information in terms of tactile diagrams is one alternative that has been proven useful, creating these diagrams remains a complex art that usually requires participation of a trained human. This is because to be effective tactile diagrams must be simplified; otherwise, they may be impossible to understand. Automating this process would have advantages in terms of cost, in terms of providing diagrams in a timely fashion (or even providing them at all), and in terms of independence for the intended users. In this project, which takes a first step toward this goal, the PI and her team will focus on the automatic conversion to tactile diagrams of two kinds of graphics: photographs and line drawings. To this end, a multi-step process is envisaged. First, techniques to segment images without a lot of fragmentation will be examined, modified and/or developed based on the current computer vision literature. Next, techniques will be developed for automating the simplification process, based on the manual steps as outlined by the Braille Authority of North America. Two main aspects will be considered: simplifying lines/curves in the diagram, and reducing clutter (defined as a situation where components of the graphic are too close together or so similar that they are hard to distinguish tactually). Then, automatic, metric rectification techniques will be explored, modified and/or developed to remove perspective from an image and replace it with a standard canonical view (because removing perspective seems to improve the user's ability to understand a tactile diagram). Finally, user studies will be conducted to evaluate the effectiveness of the tactile graphics created by the new automatic methods in comparison to those created manually by expert tactile diagram makers.Broader Impacts: The ultimate goal of this line of research is to provide people who are blind or visually impaired with immediate and automatic access to any desired information, and thereby to remove the current barriers due to the graphical nature of computers and visual digital media. Project outcomes will be useful not only to adult members of the target communities; they will also make it possible to provide access to children's picture books to promote development in blind and visually impaired youngsters. The project will train at least one graduate and several undergraduate students in understanding the needs of the target communities and developing assistive technology for them. Results will be disseminated through the usual scientific channels and also in presentations to relevant stakeholders including individuals who are visually impaired, their teachers and other rehabilitation professionals. Results from this work will also be presented in workshops to K-12 students at the Math, Science, Innovation Center of Richmond to foster interest in STEM fields.