Sparse Dictionary Pruning for Video Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video-based anomaly detection in surveillance systems faces challenges such as detecting abnormal patterns in complex scenarios with occlusions and clutter, developing computationally simple algorithms, and lacking standardized data sets for rare anomalous events.
Innovation Solution
The implementation of sparse reconstruction techniques, which involve constructing a dictionary of normal events, generating sparse codes, and optimizing it to reduce redundancy and improve discriminability, allowing for efficient detection of anomalies in video footage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a comprehensive dictionary of normal events is constructed to improve detection accuracy, then detection precision is improved, but computational complexity increases
Solution Approach 1:
The patent extracts and removes redundant dictionary columns from the comprehensive dictionary of normal events. By analyzing sparse codes and identifying redundant atoms (columns) in the dictionary, the system eliminates unnecessary computational elements while preserving the essential features needed for anomaly detection, thus reducing computational complexity without significantly compromising detection precision
Solution Approach 2:
The patent optimizes dictionary parameters by adjusting the number of dictionary atoms and selecting optimal subsets. Through parameter optimization, the system finds the right balance between dictionary size and detection performance, achieving high precision with reduced computational burden by changing the structural parameters of the dictionary
2Productivity
If real-time anomaly detection is implemented to improve productivity, then detection speed is improved, but measurement precision deteriorates
Solution Approach 1:
The patent extracts only the most discriminative and essential features from video data for real-time processing. By removing redundant information and focusing on key anomaly indicators, the system achieves fast detection speed while maintaining high precision through selective feature extraction rather than processing all available data
Solution Approach 2:
The patent performs preliminary dictionary construction and optimization during the training phase before real-time detection. By pre-processing and preparing the dictionary structure in advance, the system reduces the computational burden during real-time operation, enabling both high speed and high precision detection when actual anomalies need to be identified
Data Source
AI summary
Methods, systems, and processor-readable media for pruning a training dictionary for use in detecting anomalous events from surveillance video. Training samples can be received, which correspond to normal events. A dictionary can then be constructed, which includes two or more classes of normal events from the training samples. Sparse codes are then generated for selected training samples with respect to the dictionary derived from the two or more classes of normal events. The size of the dictionary can then be reduced by removing redundant dictionary columns from the dictionary via analysis of the sparse codes. The dictionary is then optimized to yield a low reconstruction error and a high-interclass discriminability.