Development of a Wearable Smart Guide Device for the Blind
Yi Li, Chenfeng Xiao, Juanjuan Gu, Rong‐Chao Peng · 2021
We developed a wearable smart guide device that can recognize the traffic lights and the crosswalk to help the blind to cross the road. The device uses a camera to real-time capture the road environment images and transmit them to Raspberry Pi, which runs a Python algorithm to detect the traffic lights and the crosswalk. The algorithm, on the one hand, partitions the captured image with a threshold in HSV color space, and then calculates its similarity with a pre-set template of the traffic light using template matching, and finally determines whether the traffic light exists and figures out its position if it exists; on the other hand, detects the edges in the captured image using Canny algorithm, and then finds out the straight lines in the edges using Hough Transform, and finally determines whether the crosswalk exists and figures out its position if it exists. The results of the recognized traffic lights or crosswalk are then converted to speeches through text-to-speech engines in Baidu Artificial Intelligence Cloud, and played to the user via a loudspeaker or Bluetooth earphones. After tested in different real situations, the developed device ran well and successfully recognized the traffic lights and the crosswalks. We think it is useful to bring convenience for the blind people walking outside.