Real-Time Image Segmentation Using Visible Light and Distance Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies for real-time digital image segmentation in videoconferencing and radiocommunications are inefficient, requiring complex and expensive setups, and often necessitate specific backgrounds, which limits their use in casual settings and increases bandwidth usage by transmitting unnecessary image data.
Innovation Solution
A method that uses a combination of visible and invisible image capture devices to segment the user's face from the background in real-time, allowing for reduced bandwidth usage and flexible deployment without specific background requirements, by capturing and processing digital images with synchronized focal lengths and image refresh rates, and adjusting distance thresholds to follow the user's movement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If infrared projection and analysis systems are used to distinguish people from objects, then measurement precision is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent replaces complex infrared projection and analysis systems with a simple visible light camera system. Instead of using electromagnetic wave projections and complex analysis, the invention uses standard visible light imaging combined with simple image processing algorithms to distinguish users from backgrounds, thereby reducing device complexity while maintaining measurement precision
Solution Approach 2:
The patent substitutes expensive, complex infrared systems with inexpensive, readily available visible light cameras. The solution uses common consumer-grade camera hardware and software-based processing rather than specialized expensive equipment, making the system accessible for real-time videoconferencing applications
2Measurement precision
If depth images and complex processing operations are used to locate faces, then measurement precision is improved, but productivity decreases due to processing time
Solution Approach 1:
The patent extracts only the essential features needed for user segmentation from the full image data. Instead of performing complex depth image processing and 3D analysis to locate faces, the system extracts key visual features using simple image processing operations that can be executed in real-time, maintaining accuracy while improving processing speed
Solution Approach 2:
The patent applies partial processing by focusing only on the necessary image regions and features for user segmentation. Rather than processing entire depth images or performing exhaustive face location algorithms, the system processes only the relevant portions of visible light images, achieving real-time performance while maintaining sufficient measurement precision
3Manufacturing precision
If specific background decoration is required for segmentation, then manufacturing precision is improved, but adaptability decreases
Solution Approach 1:
The patent creates a universal segmentation system that works with any background type. The visible light camera-based approach combined with image processing algorithms can handle diverse backgrounds (solid colors, patterns, complex scenes) without requiring specific background decorations, making the system adaptable to any environment while maintaining segmentation accuracy
Solution Approach 2:
Instead of requiring the environment to conform to system requirements (specific background decoration), the patent inverts the approach by making the system adaptable to any environment. The image processing algorithms are designed to work with diverse backgrounds, eliminating the need for users to modify their surroundings, thus improving adaptability while maintaining segmentation precision
4Measurement precision
If full image data is transmitted, then measurement precision is improved, but loss of energy increases due to bandwidth consumption
Solution Approach 1:
The patent extracts and transmits only the segmented user portion of the image rather than the complete image data. By isolating the user from the background through image processing, the system transmits a reduced data set that maintains essential image quality for communication purposes while significantly reducing bandwidth consumption and energy loss
Solution Approach 2:
The patent transmits partial image data containing only the user region of interest rather than the full image. This partial transmission approach maintains sufficient image quality for videoconferencing while reducing the data volume and energy required for transmission, achieving an optimal balance between quality and energy efficiency
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
Enables efficient real-time transmission of segmented images, reducing bandwidth requirements and allowing for flexible deployment in any setting, while accurately distinguishing the user from their environment, even when objects are between the user and the camera.
Implementation Method 1
an invisible image capture device for capturing a reflected image of the zone of observation by means of electromagnetic waves in a domain invisible to the human eye
Implementation Method 2
electromagnetic waves in a domain invisible to the human eye, in particular in the infrared domain
Data Source
Figure 1
Figure 2
AI summary
To segment in real time a first digital image (RI) of an observation area (ZO) captured by a first device (CIV) and transmitted to a computing device (DSI), a second digital image (INC), in which each pixel is associated with a quantity such that a distance (DP) relative to a point in the observation area is captured and transmitted by a second device (CII) to the computing device. An image fusion module (FIC) establishes a correspondence between pixels of the first image and pixels of the second image and selects the pixels of the first image corresponding to the associated pixels of the second image based on quantities associated with those pixels. Only the selected pixels of the first image are displayed to form a segmented image (IS).