Oblique Security Camera System for POS Facial Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional 'look-down' video security cameras positioned directly overhead of Point of Sale (POS) terminals have limitations in capturing facial recognition and height information due to their perspective, which is crucial for loss prevention and law enforcement, as they can only capture the tops of individuals' heads and shoulders.
Innovation Solution
A security video system with cameras positioned outside the POS area, utilizing a security video analytics system to generate image data and analyze it by drawing customer regions with vertical edges, creating bounding boxes around individuals, and comparing these to infer proximity to the POS area, overcoming perspective issues by using trapezoidal-shaped customer regions to account for three-dimensional space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If look-down cameras are positioned directly overhead of POS terminals, then customer movement near POS terminals can be monitored, but facial recognition and height determination become limited or impossible
Solution Approach 1:
The patent transitions from a vertical overhead viewing angle to an external oblique viewing angle, adding a horizontal dimension to the camera positioning. This dimensional change allows the camera to capture facial features and full-body height information while still monitoring customer movement near the POS terminal, resolving the contradiction between measurement precision and device complexity.
Solution Approach 2:
The patent introduces customer regions as virtual boundaries defined by vertical edges that serve as intermediaries between the camera's external viewpoint and the POS terminal area. These regions enable the system to accurately determine customer proximity and movement while maintaining the external camera positioning that provides facial and height information.
2Area of stationary object
If security camera is positioned outside the POS area, then wider field of view and enhanced perspective are achieved, but perspective issues arise that make it difficult to accurately determine customer proximity
Solution Approach 1:
The patent segments the monitored area into distinct customer regions defined by vertical edges. This segmentation creates clear, measurable zones that simplify the determination of customer proximity despite the external camera positioning and resulting perspective distortion. The segmented regions provide reference frames that maintain measurement accuracy.
Solution Approach 2:
The patent changes the geometric parameters of the customer regions to account for perspective distortion. By defining regions with vertical edges and adjusting their positional parameters in the image coordinate system, the system compensates for the oblique viewing angle and maintains accurate proximity detection while preserving the wider field of view from external positioning.
Data Source
AI summary
A system and method for determining proximity of individuals to specific regions of interest such as Point of Sale (“POS”) terminals within a POS monitored area of a scene of video data captured from security cameras as part of a networked security system, when the security cameras are located outside the POS monitored area, mounted on a ceiling or other high location and pointed at the POS monitored area. With the help of customer regions drawn in an abstract layer whose areas coincide with expected locations of individuals near POS terminals in the video data, and video analytics elements such as bounding boxes generated around individuals in the video data, the system can perform live and forensic analysis of the video data to infer information such as the proximity of individuals to POS terminals and the relative height of an individual compared to their expected height within the scene of video data.