Temporal Noise Reduction Filter With Adaptive Frame Blending

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing temporal noise reduction techniques in low light video capture often result in noisy images due to overreliance on previous frames with low variance, leading to undesirable noise and slow convergence.

Innovation Solution

A filter incorporating a Kalman gain filter with Sigmoid interpolation is used to account for noise variance and temporal differences between frames, determining optimal blending ratios to reduce noise effectively and quickly.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a high blending ratio is used when there is low temporal difference between frames, then temporal noise reduction is improved, but noise is introduced due to overreliance on previous frame

Engineering Contradiction:
Improvetemporal noise reductionVSAvoidnoise
Core Design Contradiction:
ReliabilityVSObject-generated harmful factors

Solution Approach 1:

The blending ratio is made dynamic and adaptive rather than fixed. The system automatically adjusts the blending ratio based on the measured temporal difference between frames, using high blending ratios when temporal difference is low and low blending ratios when temporal difference is high, thereby optimizing noise reduction while avoiding overreliance on previous frames

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system measures the temporal difference between current and previous frames as feedback, then uses this measurement to adjust the blending ratio accordingly. This closed-loop approach ensures that the blending ratio adapts to actual temporal variations, preventing both excessive smoothing and insufficient noise reduction

Inventive Principle:
Principle #23Feedback

2Object-generated harmful factors

If a low blending ratio is used when there is high temporal difference between frames, then noise from previous frame is reduced, but temporal noise reduction effectiveness decreases

Engineering Contradiction:
Improvenoise from previous frameVSAvoidtemporal noise reduction effectiveness
Core Design Contradiction:
Object-generated harmful factorsVSReliability

Solution Approach 1:

The blending ratio dynamically adapts to the temporal difference magnitude. When temporal difference is high, the system automatically reduces the blending ratio to prevent propagating noise from previous frames, while when temporal difference is low, it increases the blending ratio to maximize noise reduction effectiveness

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system uses temporal difference measurement as feedback to control the blending ratio. This feedback mechanism ensures that the blending ratio is optimally adjusted based on actual frame-to-frame variations, preventing noise propagation when temporal difference is high while maintaining effective noise reduction when temporal difference is low

Inventive Principle:
Principle #23Feedback

3Object-generated harmful factors

If traditional temporal noise reduction is used, then noise is reduced, but convergence is slow

Engineering Contradiction:
ImprovenoiseVSAvoidconvergence time
Core Design Contradiction:
Object-generated harmful factorsVSLoss of time

Solution Approach 1:

The system performs preliminary measurement of temporal difference between frames before applying the blending operation. This preliminary action allows the system to pre-determine the optimal blending ratio, enabling faster convergence by avoiding iterative adjustment and directly applying the appropriate blending weight from the start

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system changes the blending ratio parameter based on measured temporal difference. By adjusting this critical parameter dynamically rather than using a fixed value, the system achieves faster convergence to optimal noise reduction results, reducing the time required to reach effective noise suppression

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4189634B1Filter for temporal noise reduction
Publication Date: 2025.12.03 GOOGLE LLC
  • EP4189634B1 patent drawingFigure 1
  • EP4189634B1 patent drawingFigure 2A~2B
  • EP4189634B1 patent drawingFigure 3

AI summary

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for using a filter for temporal noise reduction. In some implementations, image data for a series of frames, including a first input frame followed by a second input frame, of a video is obtained. A first output frame resulting from noise reduction processing for the first input frame and a measure of variance associated with a portion of the first output frame is obtained. An interpolation setting for noise reduction processing of a portion of the second input frame is determined. A second output frame is generated by interpolating the portion of the second input frame with the corresponding portion of the first output frame.