Supervised Learning Rank Algorithm of Congestion Degree in Traffic Scenes

For automatic ananlysis of traffic scene attributes (‘congestion’,’ average speed’), traffic scene congestion degree rank calculation model is proposed based on supervised learning. Using supervised learning ideas,we learn a ranking function per attribute (‘congestion’,’average speed’). For traffic congestion degree rank model, we extract Gist feature of each frame training images, however, for average speed degree rank model, firstly, we extract video motion information and then extract Gist feature, finally, we introduce modified Ranking SVM projection model to get rank model of traffic congestion degree and speed degree. Experimental results on three types of databases show that the proposed rank model has more accuracy and stability.