Applications of Computational Verbs to Effective and Realtime Image Understanding

An effective and realtime image understanding algorithmmust meet the challenges of ever-growing image databases,the varieties of features to be understood and the limited humanresources for developing this algorithm. In this paper, the reasonsthat mainstream image understanding algorithms could not meetthese challenges were presented. One effective and realtime imageunderstanding platform based on physical linguistic(PL) networkand cognitive features was given. The effective aspect of thePL network-based image understanding engine stems from theinfinite flexibility of human natural languages when cognitivefeatures are grouped and modelled. The realtime aspect of thePL network-based image understanding engine originates fromthe much faster processing speed for computational nouns andcomputational verbs than the processing speed for raw imagedata.The applications of this new effective and realtime imageunderstanding platform to image search engine, porn detection/filtering and many other image understanding tasks werecompared with the conventional approaches. The results showthat the new image understanding platform is much moreflexible, accurate, user-friendly and less resource consuming.