People Flow Analysis Using Image Segmentation for High-Resolution Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing people flow analysis methods struggle with real-time analysis of high-resolution images due to increased image size, leading to slow processing and inability to detect abrupt changes, and lack sufficient accuracy in distinguishing local patterns of people flow.
Innovation Solution
An image processing apparatus that sets a measurement area within an input image, divides it into smaller images, estimates flow distribution, and calculates the number of targets passing through the area, using techniques like deep learning and probabilistic modeling to enhance analysis accuracy and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire high-resolution image is used for people flow analysis, then the analysis accuracy is improved, but the processing speed deteriorates
Solution Approach 1:
The patent divides the high-resolution image into multiple small images (sub-images) and performs people flow analysis on each small image independently. This segmentation approach maintains analysis accuracy by preserving local flow patterns while significantly reducing processing time compared to analyzing the entire high-resolution image at once.
2Measurement precision
If the optical flow of all pixels is averaged to calculate moving speed, then the general tendency of people flow is accurately obtained, but the local people flow change cannot be tracked
Solution Approach 1:
Instead of averaging optical flow across the entire image, the patent segments the image into multiple small images and calculates flow distribution for each segment separately. This preserves local flow information and enables tracking of local people flow changes while still providing overall flow trends.
Solution Approach 2:
The patent applies different analysis approaches to different regions by examining flow distribution in each small image separately. This allows local flow patterns to be captured with high fidelity while maintaining the ability to derive general tendencies from aggregated results.
3Measurement precision
If a plurality of patterns of people flow are distinguished locally, then the analysis accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent simplifies the complexity of distinguishing multiple people flow patterns by dividing the image into small images and analyzing each independently. This segmentation reduces the computational complexity compared to analyzing the entire high-resolution image while maintaining the ability to distinguish local flow patterns.
Data Source
AI summary
An image processing apparatus includes a measurement area setting unit that sets a measurement area in an input image, a small image setting unit that sets a small image in the input image based on the measurement area, a first estimation unit that estimates a flow distribution of a target in the small image, and a second estimation unit that estimates the number of targets to pass through the measurement area based on the flow distribution in the small image.