Video Congestion Analysis Using Motion Trajectories and Crowd State
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing congestion estimation systems fail to accurately determine the congestion degree in terms of the number of people per unit area, leading to inadequate assessment of congestion states, particularly in areas with large gatherings where movement is difficult.
Innovation Solution
A congestion state visualization device and method that utilizes video image photographing devices, a congestion analysis device, and a display device to analyze moving images, extract feature points and trajectories, classify persons, determine motion trajectories, and calculate congestion degrees based on the number and state of individuals, outputting congestion information through heat maps and icons.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the congestion degree is determined only by detecting the presence of people using texture information, then the detection process is simple, but the measurement precision of congestion state is insufficient
Solution Approach 1:
The patent changes the parameter from binary presence detection to quantitative measurement by extracting the number of people per unit area as a key parameter. This is achieved by dividing the detection area into multiple regions and counting people in each region, transforming the congestion assessment from qualitative to quantitative measurement.
Solution Approach 2:
The detection area is segmented into multiple divided areas, and people are counted separately in each area. This segmentation allows for more precise localization and quantification of congestion levels at different spatial positions, improving overall measurement precision.
2Measurement precision
If the area is divided into multiple regions for detailed analysis, then the measurement precision of congestion distribution is improved, but the device complexity increases
Solution Approach 1:
The monitoring area is automatically divided into multiple grid regions, with people counted in each region independently. This segmentation enables precise mapping of congestion distribution across different areas while using systematic processing methods to manage the complexity of multi-region analysis.
Solution Approach 2:
Different congestion levels in divided areas are represented by different colors in the output image, with color intensity or hue indicating the degree of congestion. This visual encoding simplifies the interpretation of complex multi-region data by transforming numerical counts into intuitive visual information.
3Reliability
If only the presence of people is detected without considering movement state, then the detection process is simple, but the reliability of congestion assessment is insufficient
Solution Approach 1:
The system analyzes the dynamic state of people by detecting motion in video sequences. By examining changes in person positions across multiple frames, the system distinguishes between moving and stationary individuals, providing more reliable congestion assessment that reflects actual crowd dynamics rather than just static presence.
Solution Approach 2:
The system uses feedback from motion detection to refine congestion assessment. By continuously monitoring movement patterns and comparing them against congestion criteria, the system adjusts its assessment to account for the dynamic nature of crowds, improving reliability through iterative evaluation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention suitably determines how is congestion in an area. A congestion analysis device (4) includes: a person classifying unit (43) that is configured to determine, based on a moving image in which persons are photographed, the number of the persons; a motion trajectory determination unit (44) that is configured to determine a motion trajectory of each of the persons; and a congestion information analysis unit (45) that is configured to determine a state of each of the persons based on the motion trajectory thereof photographed in the moving image and also determine a congestion degree in accordance with the number of the persons and the state of each of the persons.