Spatio-Temporal Clustering for Abnormal Stay Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining abnormal stay states of objects in specific locations, such as near prohibited points of interest, face challenges in accuracy due to reliance on image data which is limited by cost and error-prone feature recognition, and lack of temporal data consideration in clustering algorithms like DBSCAN, leading to reduced recall rates and accuracy.
Innovation Solution
The integration of a spatio-temporal constraint clustering algorithm (ST_DBSCAN) with rule-based determination using temporal displacement data to identify abnormal stay points, improving accuracy and recall rates by processing and preprocessing displacement data to determine valid stay points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image data and DBSCAN clustering algorithm are used to determine abnormal stay states, then the system can identify stay states of objects, but the accuracy and recall rate are reduced due to limited image data quality and lack of temporal data consideration
Solution Approach 1:
The patent transitions from purely spatial clustering (DBSCAN on coordinates) to spatio-temporal clustering by adding the time dimension. The ST_DBSCAN algorithm processes moving point sequences that include both spatial coordinates and timestamp information, enabling the system to distinguish between transient stops and actual abnormal stays by analyzing temporal patterns alongside spatial proximity to prohibited points of interest.
Solution Approach 2:
The patent introduces new parameters for abnormal stay determination: minimum stay duration threshold and maximum distance threshold from prohibited POI. These parameter changes allow the system to filter out false positives by requiring both temporal duration and spatial proximity conditions to be met simultaneously, thereby improving both accuracy and recall rate compared to simple spatial clustering.
2Productivity
If simple spatial clustering is used without temporal data, then the processing is computationally simpler, but the recall rate decreases due to inability to distinguish transient stops from abnormal stays
Solution Approach 1:
The patent extends the clustering algorithm from 2D spatial coordinates to 3D spatio-temporal data by incorporating timestamp information into the moving point sequences. This dimensional extension enables the algorithm to identify temporal patterns such as prolonged stays versus transient stops, significantly improving recall rate while maintaining computational feasibility through efficient sequence processing.
3Measurement precision
If image recognition methods are used to identify objects near prohibited POI, then visual confirmation is obtained, but the system becomes more expensive and error-prone due to feature recognition limitations
Solution Approach 1:
The patent replaces image recognition and visual confirmation systems with computational methods based on spatio-temporal data processing. Instead of using cameras and image analysis algorithms, the system uses structured moving point sequences with timestamp and coordinate data processed through ST_DBSCAN clustering, eliminating the need for expensive and error-prone image recognition infrastructure while maintaining or improving identification accuracy.
Data Source
AI summary
A method of determining a state of a target object, an electronic device, and a storage medium, relate to fields of a computer technology, cloud computing and Internet of things, and apply to smart cities. The method includes: receiving a transmitted first moving point sequence for the target object, the first moving point sequence including a plurality of target moving point elements, and each target moving point element containing a timestamp information and a displacement information that indicate a stay state of the target object; determining, from the first moving point sequence, a target stay point of the target object, according to the timestamp information and the displacement information; and determining that the state of the target object at the target stay point is an abnormal stay state, in response to a distance between the target stay point and a first preset position being less than a first preset threshold.