| Summary: | 碩士 === 國立中央大學 === 資訊工程研究所 === 93 === The most pervasive travel-aid for the blind is the White Cane, which only allows them to ensure that the path is clear and to locate any possible obstacle by using it to touch the ground and objects. The rest of the world beyond reach remains unknown to them. The other option for a blind person to get around is by the help of a guide dog. However, the present population of the blind in Taiwan seriously outnumbers the existing guide dogs. A guide dog training program is very selective as to the dogs to be trained, and it costs a lot of time and money to train while the lifespan of a guide dog is rather limited. As a result, the use of guide dogs is not very popular. Therefore, it is crucial to develop an easy-to-use and cost-efficient electronic guiding system to improve the lives of the visually disabled. At present, a large portion of existing guiding aids uses ultrasonic to detect objects. But instead of actively detecting the surroundings, users need to detect objects by themselves and then avoid whatever possible obstacles. They are not able to choose the path according to their will since this kind of tools only provides passive guidance. In the thesis, we proposed a travel-guide which requires a low-cost web camera and an accelerator circuit. First, we use a segmentation algorithm to segment the floor region from the image captured by the web camera. By the use of the tilting angle measured from the accelerator circuit and a transformation formula we are able to gather information about the width and length of the floor. These two kinds of information are the most critical information for the blind while walking. The information answers their most urgent question about “how wide is the road and how far can I go?” This research explores the ways to help the blind to navigate more actively and be able to move as if they can see. Experiments were conducted to test the performance of the proposed travel-aid.
|
|---|
