基于记忆单元与多尺度结构相似性的异常检测  

Anomaly detection based on the memory unit and multi-scale structural similarity

作　　者：王婷[1] 宣士斌[2,3] 付孟丹 周建亭 WANG Ting;XUAN Shibin;FU Mengdan;ZHOU Jianting(College of Electronic Information,Guangxi Minzu University,Nanning 530006,Guangxi,China;College of Artificial Intelligence,Guangxi Minzu University,Nanning 530006,Guangxi,China;Guangxi Key Laboratory of Hybrid Computation and IC Design and Analysis,Nanning 530006,Guangxi,China)

机构地区：[1]广西民族大学电子信息学院,广西南宁530006 [2]广西民族大学人工智能学院,广西南宁530006 [3]广西混杂计算与集成电路设计分析重点实验室,广西南宁530006

出　　处：《微电子学与计算机》2023年第8期28-36,共9页Microelectronics & Computer

基　　金：国家自然科学基金资助项目(61866003);广西民族大学研究生教育创新计划项目(gxun-chxs2021063)。

摘　　要：针对基于记忆单元的自编码器模型(Dynamic Prototype Unit Model,DPU)在检测视频异常时没有充分利用多层次特征、未考虑异常与正常事件间的结构性差异的问题,提出融合多尺度记忆模块和多尺度结构相似性的异常检测模型.新模型构建了多尺度记忆模块(Multi Scale Memory Module),利用不同尺度空间的记忆单元对编码层特征进行编码,并将编码结果与解码层特征拼接,既能保留网络的浅层细节信息,又能促进正常模式的多样性.为了约束对正常事件中结构信息的学习,组合多尺度结构相似性(Multi Scale Structure Similarity Index,MS-SSIM)误差与误差作为目标函数,使预测视频中的事件结构更接近正常事件,提高视频中异常事件的预测误差.在标准数据集UCSD Ped1、UCSD Ped2和Avenue数据集上的实验结果表明,提出模型的帧级AUC比原模型分别提高了0.8%、3.4%和1.0%,帧率达到142.9 fps.In order to solve the problem that the dynamic prototype unit model based on memory unit does not make full use of mmulti-level features and does not consider the structural differences between abnormal and normal events when detecting video anomalies,a anomaly detection model combining mmulti-scale memory module and multi-scale structural similarity is proposed.The new model constructs a multi-scale memory Module,which uses memory units of different scale space to encode the features of the encoding layer,and concatenates the encoding results with the features of the decoding layer,which can not only preserve the shallow details of the network,but also promote the diversity of normal patterns.In arder to constrain the learning of structural infomnation in nomal events,the multi scale structure similanity index error and L error are combined as objective fiunctions to make the event structure in the predicted video closer to the normal event,and improve the prediction ernror of abnormal events in the video.The experimental results on the standard datasets UCSD Ped1,UCSD Ped2 and Avenue show that the framne level AUC of the proposed model is improved by 0.8%,3.4%and 1.0%compared with the original model,respectively.And the frame rate reaches 142.9 FPS.

关 键 词：视频异常检测 记忆单元 多尺度结构相似性 自编码器 MS-SSIM 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]