Research of Active Perception Model Detecting Vehicle's Anomalous Motions
LU Xingya;YANG Feng;JI Wei;Police Data Center,Xuzhou PSB;Domestic Security Brigade,Pizhou PSB;
In this Big data era, data have become essential resources. The law enforcement organizations have realized that data play a crucial role in public security. The police departments hope that their vast amount of data generated by surveillance cameras can make effective contributions to social order, public safety, and criminal investigation. The use of these data yet involves great manual power in searching, comparison and recognition, hence holds a low intelligent level, but a high time cost. Based on the research on patterns of vehicles, this paper proposes a active perception model detecting vehicle's anomalous motions. With the implementation of deep neural network training and 2-D plane clustering algorithms, the model improves the efficiency of the police daily work, and makes substantial achievements in real-world applications.
【CateGory Index】： D631.5;U495