郑州大学学报（理学版）2025，Vol.57Issue(5)：39-45,7.DOI:10.13705/j.issn.1671-6841.2024029
基于多尺度量化特征的视频异常行为检测算法
Video Abnormal Behavior Detection Method Based on Multi-scale Quantization Features
摘要
Abstract
Abnormal behavior detection in video had significant application value in the field of surveil-lance and security.To address the issue of abnormal information generalization caused by skip connec-tions between the encoder and decoder in autoencoder models in generating video frames,an algorithm for video abnormal behavior detection based on multi-scale quantized features was proposed.Firstly,the en-coder was trained to learn normal frames and performed vector quantization in a hierarchical manner,while the decoder generated video frames based on the quantized features,avoiding direct information transmission between the encoder and decoder,to significantly reduce the impact of generalization,and to improve the quality of frame generation.Secondly,a pyramid deformation module was utilized to measure the diversity of the generated frames,to calculate the deformation between the generated frames and the original frames to measure the severity of the abnormality.Finally,the anomaly score was obtained by fu-sing the reconstruction error of the generated frames.The abnormal detection performance of the algorithm was tested on public datasets,and the experimental results showed that the AUC value of the proposed al-gorithm was higher than that of similar algorithms.关键词
视频异常检测/多尺度/矢量量化/变分自编码器Key words
video anomaly detection/multi-scale/vector quantization/variational autoencoder分类
信息技术与安全科学引用本文复制引用
马建红,王亚辉,靳岩,卫权岗..基于多尺度量化特征的视频异常行为检测算法[J].郑州大学学报（理学版）,2025,57(5):39-45,7.基金项目
国家重点研发计划项目(2020YFB171240) （2020YFB171240）
郑州市协同创新重大专项(20XTZX06013) （20XTZX06013）
