Video anomaly detection based on locality sensitive hashing filters

摘要: In this paper, we propose a novel anomaly detection approach based on Locality Sensitive Hashing Filters (LSHF), which hashes normal activities into multiple feature buckets with (LSH) functions to filter out abnormal activities. An online updating procedure is also introduced the framework of LSHF for adapting changes video scenes. Furthermore, develop new evaluation function evaluate hash map and employ Particle Swarm Optimization (PSO) method search optimal functions, improves efficiency accuracy proposed method. Experimental results datasets demonstrate that algorithm capable localizing various in real world surveillance videos outperforms state-of-the-art methods. HighlightsWe present locality sensitive hashing filters detection.Normal are hashed by build filters.Abnormality test sample estimated response its nearest bucket.Online mechanism increase adaptability scene changes.Searching accuracy.Our performs favorably against previous algorithms.