In order to assist people with visually degraded or damaged to obtain the information from the dynamic pedestrian traffic signals at the intersection, this system captured the image in front of the pedestrian through the camera, and analyzed the content of the image to recognize the dynamic pedestrian traffic signal (traffic lights with the countdown seconds) at the intersection. Through the Haar-like with HSV color correction in the experiment, the average accuracy of the recognition result was 94.6%, and the average recall rate was 81.6%. In addition, as for the digit identification of the countdown seconds, the coordinates of the seconds were estimated first, the digit image was centered and then divided. Finally, the digit recognition was performed and then converted back to the number of seconds. The average precision of the recognition result was 95.2%, and the average recall rate was 77.2%.
CITATION STYLE
Wu, C. C., & Hsug, Y. C. (2020). A research on constructing the recognition system for the dynamic pedestrian traffic signals through machine vision. In Advances in Intelligent Systems and Computing (Vol. 1036, pp. 493–501). Springer Verlag. https://doi.org/10.1007/978-3-030-29029-0_47
Mendeley helps you to discover research relevant for your work.