Dual-Angle People Counting via Covering Ratio Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current pedestrian counting methods, such as WiFi signal analysis, GPS, and infrared sensors, face challenges in accuracy and complexity, while camera-based systems struggle with noise and interference, necessitating a more efficient and accurate approach for counting people in specific areas.
Innovation Solution
A method and system utilizing dual-angle image capture and analysis, where top-down and side-angle images are processed to detect human bodies, calculate covering ratios, and estimate the number of people within a shot zone, with a front-end controlling unit and back-end deep learning for enhanced accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If camera-based counting method is used, then the ability to analyze specific area and filter noise is improved, but the accuracy decreases when objects are covered or interfered
Solution Approach 1:
The patent transitions from single-angle camera viewing to multi-angle (top-down and side) camera arrangement. By adding the vertical dimension (top-down view) combined with side view, the system can distinguish covered objects through their shadows and projections, resolving the accuracy loss when objects are covered in single-view imaging
2Device complexity
If interrupted sensors are used, then the counting process is simplified, but the accuracy decreases because connected articles passing through together cannot be identified
Solution Approach 1:
The patent segments the counting process into multiple independent stages: shadow detection from top-down view, projection matching from side view, and sequential identification. This segmentation allows the system to handle connected articles by analyzing their spatial relationships across different views, identifying individual objects within groups
3Area of stationary object
If GPS-based counting method is used, then the coverage area is expanded, but the calculation complexity and data requirements increase significantly
Solution Approach 1:
The patent extracts only the essential geometric features (shadows and projections) from the imaging data, discarding unnecessary image details. This extraction approach maintains expanded coverage capability while dramatically reducing calculation complexity by working with simplified geometric representations rather than full GPS coordinate sets
Data Source
AI summary
A method for analyzing a number of people includes an image shooting step and a front-end analyzing step. In the image shooting step, a first image and a second image are obtained. In the front-end analyzing step, a foreground object analysis is operated, and a plurality of foreground objects located at a region of interest in the first image are obtained. A human body detection is operated, and at least one human body and a location thereof of the second image are obtained. An intersection analysis is operated, and the location of the human body is matched to the first image. A number of people estimation is operated to estimate the number of the people according to the first covering ratio, a number of the human body, and a second covering ratio of all the foreground objects to the region of interest.