Video Encoder Latency Reduction via Dynamic Quantization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video encoding technologies face challenges in reducing latency while maintaining acceptable video quality, particularly in applications requiring low latency video transmission over long distances, such as visual aids for vehicle drivers.

Innovation Solution

A method and device that adapt the quantization parameter (QP) in real-time to limit frame sizes, using a rate control mechanism to ensure frame sizes do not exceed processing capabilities on the receive side, thereby reducing latency by dynamically adjusting the compression rate during video encoding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If video encoding is performed with high compression to reduce transmission time, then latency is reduced, but video quality deteriorates

Engineering Contradiction:
ImprovelatencyVSAvoidvideo quality
Core Design Contradiction:
Loss of timeVSManufacturing precision

Solution Approach 1:

The patent implements dynamic rate control that continuously adapts the quantization parameter based on real-time buffer status and scene complexity. The rate control mechanism dynamically adjusts compression strength between frames, using lower compression (higher QP) when the buffer is near capacity to prevent overflow, and higher compression (lower QP) when buffer space is available, thereby reducing latency while maintaining video quality within acceptable bounds

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the quantization parameter (QP) dynamically during encoding based on buffer occupancy and scene characteristics. By adjusting QP values in real-time rather than using a fixed compression rate, the system can control the balance between latency reduction and video quality preservation, allowing flexible adaptation to different transmission conditions and buffer states

Inventive Principle:
Principle #35Parameter changes

2Loss of time

If frame size is reduced to speed up transmission, then latency is reduced, but video quality deteriorates

Engineering Contradiction:
Improvetransmission timeVSAvoidvideo quality
Core Design Contradiction:
Loss of timeVSManufacturing precision

Solution Approach 1:

The patent performs preliminary estimation of the current frame size before actual encoding based on previous frame characteristics and scene complexity. This allows the rate control mechanism to pre-adjust the quantization parameter to achieve the target frame size, preventing buffer overflow while maintaining optimal video quality. The preliminary sizing action enables proactive latency reduction without sacrificing quality

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements a feedback loop where the actual frame size is measured after encoding and compared against the target frame size. Based on this feedback, the quantization parameter is adjusted for subsequent frames to maintain the desired frame size. This closed-loop control ensures that transmission time is optimized while video quality remains within acceptable parameters

Inventive Principle:
Principle #23Feedback

3Speed

If quantization parameter is increased to reduce frame size, then transmission speed increases, but video quality deteriorates

Engineering Contradiction:
Improvetransmission speedVSAvoidvideo quality
Core Design Contradiction:
SpeedVSManufacturing precision

Solution Approach 1:

The patent dynamically adjusts the quantization parameter based on real-time buffer status and scene complexity rather than using a fixed high QP value. When buffer occupancy is low and scenes are simple, lower QP values are used to maintain video quality. When buffer occupancy is high or scenes are complex, higher QP values are used to reduce frame size and increase transmission speed, thereby dynamically optimizing the trade-off between speed and quality

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP2985997B1Low latency video encoder
Publication Date: 2020.11.25 ALLEGRO DVT
  • EP2985997B1 patent drawingFigure 1~3
  • EP2985997B1 patent drawingFigure 4~5
  • EP2985997B1 patent drawingFigure 6~7A

AI summary

The invention concerns a method comprising: encoding, by a video encoder (102) and based on a first quantization factor (QP), one or more first groups of image blocks of a video frame; estimating, by a processing device, the size of an image slice comprising a plurality of further groups of image blocks to be encoded; and determining, based on the estimated size of the image slice, a second quantization factor (QP) to be applied to one or more of the further groups.