Research for the improvement of traffic accident detection system at an intersection applied technique of video detection and sound detection

Existing traffic accident automatic recording device, which is excessively sensitive to environmental influence, tends to deteriorate the detection rate and valid acquisition rate. In an effort to resolve such problems, a fusion algorithm using traffic accident detection algorithm based on an analysis of image detection technology-based objects recognition, pixel accumulation and fluid approximate algorithm were developed. And the field test was conducted in a bid to prove the performance of fusion algorithm. Consequently, while the detection rate with existing image detection was 50% of traffic accident detection and 0.27% of valid event detection, the rates obtained employing improved image detection algorithm reached 100% and 1.57%, respectively, and the rates obtained using an image and soundbased fusion algorithm were proved to be 100% and 21.05%, respectively.