Foreground-Background Video Segmentation for Low Bandwidth Transmission

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

High-definition video transmission over low bandwidth connections is challenging due to the significant bandwidth requirements of HD video, especially in surveillance and wireless communication applications where bandwidth is limited, such as remote sites with only 128 Kbps capacity.

Innovation Solution

A video transmission system that identifies and processes foreground and background objects separately, transmitting foreground objects at a higher update rate (e.g., 30 frames per second) and background objects at a lower update rate (e.g., every 10 minutes), using a Gaussian mixture model to differentiate semantically significant movement from repetitive background motion, thereby reducing the overall bandwidth needed.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If HD video is transmitted over low bandwidth connections, then video quality can be maintained, but bandwidth consumption becomes excessive

Engineering Contradiction:
Improvevideo qualityVSAvoidbandwidth consumption
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The video stream is segmented into two distinct components: foreground objects and background. The background is transmitted at a reduced periodic rate (e.g., once every 10 minutes or less) while foreground objects are transmitted at full frame rate. This segmentation allows the system to maintain video quality for moving objects while dramatically reducing overall bandwidth consumption by updating the static background infrequently.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different quality levels and update rates are applied to different parts of the video stream. The foreground objects receive high-quality, full-rate transmission to maintain visual fidelity for semantically significant content. The background receives lower-quality, periodic transmission since it contains repetitive, less important information. This local differentiation optimizes the trade-off between quality and bandwidth usage.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If background is transmitted at high update rate, then background quality is maintained, but bandwidth is wasted on repetitive information

Engineering Contradiction:
Improvebackground qualityVSAvoidbandwidth waste
Core Design Contradiction:
Measurement precisionVSLoss of energy

Solution Approach 1:

The background transmission follows a periodic pattern rather than continuous high-rate transmission. The system transmits the background at intervals (e.g., every 10 minutes or less) and interpolates or holds the previous background image between updates. This periodic action eliminates bandwidth waste from transmitting repetitive, unchanged background information while maintaining acceptable background quality in the reconstructed video stream.

Inventive Principle:
Principle #19Periodic action

3Speed

If foreground objects are transmitted at full frame rate, then motion detail is preserved, but overall data volume increases

Engineering Contradiction:
Improveframe update rateVSAvoiddata volume
Core Design Contradiction:
SpeedVSQuantity of substance

Solution Approach 1:

The system extracts and separates foreground objects from the background video stream using motion detection and segmentation algorithms. By taking out the foreground objects as a distinct data stream, the system can transmit them at full frame rate to preserve motion detail and semantic information, while transmitting the remaining background separately at a reduced rate. This extraction allows selective optimization of different video components.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP2659674B1Video coding
Publication Date: 2020.03.11 PELCO INC
  • EP2659674B1 patent drawingFigure 1
  • EP2659674B1 patent drawingFigure 2~3
  • EP2659674B1 patent drawingFigure 4

AI summary

Techniques are discussed for providing mechanisms for coding and transmitting high definition video, e.g., over low bandwidth connections. In particular, foreground-objects are identified as distinct from the background of a scene represented in a plurality of video frames received from a video source, such as a camera. In identifying foreground-objects, semantically significant and semantically insignificant movement (e.g., repetitive versus non-repetitive movement) is differentiated. Processing of the foreground-objects and background proceed at different update rates or frequencies.