Pedestrian Presence Detection System Based on Image Processing
Toncho Balbuzanov, Boris Ivanov Evstatiev · 2019
The study presents a methodology for detection of pedestrian presence at road crossings. It is based on image processing and a recognition background pattern. For each camera image is applied a filter in order to exclude pedestrians outside the waiting enclosure from affecting the system's decisions. Next, the object is converted to black and white format and is divided into sections. Each section is compared with the corresponding section from the recognition pattern pixel-by-pixel. The relative error between the two images is then compared to a critical value, and a decision is made. The analysis of the images by sections reduces the influence of potential vibrations, thus significantly increasing the probability for correct recognitions.