Real-Time People Counting via Infrared Depth Layer Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for counting objects in regional spaces, such as people or livestock, face inaccuracies due to ambient light interference and difficulties in handling large crowds, especially in open areas like parades or markets, where sensors like radar, color cameras, and ultrasonic sensors struggle to provide reliable counts.
Innovation Solution
A real-time layer scanning method using an infrared depth sensor to construct background and foreground depth maps, filter noise, and accurately count objects by employing erosion and dilation techniques to enhance accuracy and reduce background interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a color camera is used to count objects, then the masking effect is reduced and more objects can be detected simultaneously, but the system becomes sensitive to ambient light changes which causes incorrect foreground data
Solution Approach 1:
The patent introduces an infrared depth sensor as an intermediary device that captures depth information independent of ambient light conditions. This mediator translates the counting problem into the depth domain, where light interference is eliminated, while preserving the ability to distinguish objects from background through depth differentiation.
Solution Approach 2:
The patent replaces the optical-based color camera system with an infrared depth sensing system. This substitution moves from visible light detection (mechanically affected by ambient light) to infrared depth mapping (immune to ambient light), fundamentally changing the detection mechanism while maintaining object counting capability.
2Quantity of substance
If a radar sensor is used to detect moving objects, then the system can estimate moving rate and direction, but it becomes difficult to calculate the number of objects accurately when the crowd is too large
Solution Approach 1:
The patent transitions from radar's velocity/direction measurement capability to depth mapping in the spatial dimension. By capturing depth information for each pixel, the system can individually identify and count objects in crowded scenes, overcoming radar's limitation in accurately counting large numbers of closely spaced objects.
Solution Approach 2:
The patent segments the scene into individual depth contours for each object. By extracting and analyzing separate depth contours in the depth map, the system can identify and count each object individually even when many objects are present, whereas radar sensors struggle to distinguish individual objects in dense crowds.
3Object-affected harmful factors
If infrared depth sensor is used to count objects, then the system is not affected by ambient light changes, but noise filtering and object separation require complex processing steps
Solution Approach 1:
The patent changes the parameter domain from visual appearance (affected by light) to depth measurement (immune to light). This parameter transformation fundamentally eliminates ambient light interference while the subsequent processing operates on the more stable depth parameter, reducing the impact of environmental variations despite adding processing steps.
Solution Approach 2:
The patent performs preliminary depth map acquisition and background depth map construction before object counting. By pre-processing the depth information and establishing a background model, the system simplifies subsequent object extraction and counting operations, making the overall process more manageable despite the multiple processing steps required.
4Area of stationary object
If population density is calculated using aerial photographs, then large-scale areas can be estimated, but the method cannot provide real-time counting and is limited to static analysis
Solution Approach 1:
The patent replaces static aerial photograph analysis with dynamic infrared depth sensor capture. This substitution enables real-time depth map acquisition and processing, transforming the system from batch processing of static images to continuous real-time monitoring while maintaining the ability to cover large areas through the sensor's field of view.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method effectively improves object counting accuracy in regional spaces, reducing noise and ambient light interference, and accurately tracks objects even in crowded conditions, providing reliable population data for management and security purposes.
Implementation Method 1
an infrared depth sensor...shoots downward...using the infrared depth sensor to shoot N frames
Data Source
AI summary
Disclosed herein is a method for counting the number of the targets using the layer scanning method. The steps of this method includes constructing a background frame, filtering the noise of foreground frame and classifying the targets, and screening the area of targets based on layer scanning to calculate the number of targets by determining the highest positions of the respective targets. In addition, the dynamic numbers of targets are calculated using algorithm. Accordingly, the present invention is beneficial in automatically, effectively and precisely calculating the number of the targets in/out a specific area, achieving the flow control for targets and reducing artificial error upon calculation.