深度自编码与自更新稀疏组合的异常事件检测算法

PDF下载HTML

Title:

Abnormal event detection method based on deep auto-encoder and self-updating sparse combination

作者:

王倩倩, 苗夺谦, 张远健

同济大学 嵌入式系统与服务计算教育部重点实验室, 上海 201804

Author(s):

WANG Qianqian, MIAO Duoqian, ZHANG Yuanjian

Key Laboratory of Embedded System and Service Computing, Tongji University, Shanghai 201804, China

关键词:

深度学习; 稀疏组合; 自动编码器; 自更新; 异常事件检测; 卷积神经网络; 无监督学习; 稀疏学习

Keywords:

deep learning; sparse combination; auto-encoder; self-updating; abnormal event detection; convolution neural network; unsupervised learning; sparse representation

分类号:

TP391

DOI:

10.11992/tis.202007003

摘要:

基于深度学习的异常检测算法输入通常为视频帧或光流图像，检测精度和速度较低。针对上述问题，提出了一种以运动前景块为中心的卷积自动编码器和自更新稀疏组合学习(convolutional auto-encoders and self-updating sparse combination learning, CASSC)算法。首先，采用自适应混合高斯模型(gaussian mixture model, GMM)提取视频前景，并以滑动窗口的方式根据前景像素点占比过滤噪声；其次，构建3个卷积自动编码器提取运动前景块的时空特征；最后，使用自更新稀疏组合学习对特征进行重构，依据重构误差进行异常判断。实验结果表明，与现有算法相比，该方法不仅有效地提高了异常事件检测的准确性，且可以满足实时检测需求。

Abstract:

In the construction of a deep learning model for abnormal event detection, frames or optical flow are considered but the resulting accuracy and speed are not satisfactory. To address these problems, we present an algorithm based on convolutional auto-encoders and self-updating sparse combination learning, which is centered on the movement of foreground blocks. First, we use an adaptive Gaussian mixture model to extract the foreground. Using a sliding window, the foreground blocks that are moving, are filtered based on the number of foreground pixels. Three convolutional auto-encoders are then constructed to extract the temporal and spatial features of the moving foreground blocks. Lastly, self-updating sparse combination learning is applied to reconstruct the features and identify abnormal events based on the reconstruction error. The experimental results show that compared with existing algorithms, the proposed method improves the accuracy of abnormality detection and enables real-time detection.