Flexible Multi-level Regression Model for Prediction of Pedestrian Abnormal Behavior

摘要

The high incidence of heinous crime is increasing to use of CCTV. However, CCTV has been used to obtain evidence rather than crime prevention. Also it shows a weak effect about preventing crime. To solve the weak effort, we propose a Flexible Multi-level Regression (FMR) model that should estimate a dangerous situation for the pedestrian. The FMR model is tracking the behavior of between pedestrians from multiple CCTV that are located in different locations. The FMR has a prediction logic that should estimate an abnormal situation to analyze the possibility of crime by using the Regression and Apriori algorithm. The FMR model can be usefully used to prevent the crime because of an immediate response and rapid situation assessment.