Hybrid Video Filtering for Motion Artifact Reduction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video filtering technologies, such as motion adaptive spatio-temporal filters (MASTF), do not fully utilize temporal correlations in video sequences, leading to suboptimal performance in areas with motion and potential motion blur, and are limited by the accuracy of motion vector estimation, which can result in artifacts and increased computational cost.

Innovation Solution

A hybrid approach combining motion adaptive spatio-temporal filtering (MASTF) and motion compensated temporal filtering (MCTF) with a soft switching mechanism, using motion vector estimation and pixel motion error indicators to dynamically adjust filtering between temporal and spatial techniques, ensuring effective noise reduction and improved video quality across both motion and non-motion areas.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a temporal filter is applied in areas with motion, then noise reduction is improved, but motion blur is introduced

Engineering Contradiction:

Improvenoise reductionVSAvoidmotion blur

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent applies different filtering strategies to different regions of the video frame based on motion detection. In non-motion areas, temporal filtering is applied for noise reduction. In motion areas, spatial filtering is applied to avoid motion blur. This local differentiation resolves the contradiction by ensuring that the appropriate filter type is applied to each region according to its characteristics.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent dynamically switches between temporal and spatial filtering based on motion detection results. The filtering approach is not fixed but adapts in real-time based on the detected motion characteristics of each pixel or block, allowing the system to optimize between noise reduction and motion blur avoidance dynamically.

Inventive Principle:

Principle #15Dynamics

2Object-affected harmful factors

If a spatial filter is applied in non-motion areas, then motion blur is avoided, but noise reduction effectiveness is reduced

Engineering Contradiction:

Improvemotion blur avoidanceVSAvoidnoise reduction

Core Design Contradiction:

Object-affected harmful factorsVSReliability

Solution Approach 1:

The patent identifies non-motion areas where temporal filtering is most effective for noise reduction, and applies spatial filtering only in motion areas where it is needed to avoid motion blur. This localized application ensures that each filter type is used where it provides the most benefit, resolving the contradiction between noise reduction and motion blur avoidance.

Inventive Principle:

Principle #3Local quality

3Device complexity

If motion adaptive spatio-temporal filtering is used, then computational cost is reduced, but temporal correlation utilization is incomplete

Engineering Contradiction:

Improvecomputational costVSAvoidtemporal correlation utilization

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The patent merges motion adaptive spatio-temporal filtering with motion compensated temporal filtering in a hybrid approach. This combination allows the system to leverage the computational efficiency of MASTF while also utilizing the temporal correlation benefits of MCTF, thereby resolving the contradiction between computational cost and temporal correlation utilization.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent creates a multi-functional filtering system that can operate in different modes depending on the input conditions. The hybrid filter can function as a temporal filter, spatial filter, or a combination thereof, providing versatile noise reduction and motion handling capabilities while maintaining computational efficiency through adaptive operation.

Inventive Principle:

Principle #6Universality (Multi-functionality)

4Measurement precision

If motion vector estimation accuracy is improved, then filtering performance is improved, but computational cost increases

Engineering Contradiction:

Improvemotion vector estimation accuracyVSAvoidcomputational cost

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies motion compensated temporal filtering selectively only to regions where motion vectors are reliably estimated, rather than attempting to process the entire frame. This partial application approach improves filtering performance in motion areas while avoiding the excessive computational cost of processing all pixels with potentially inaccurate motion vectors.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS8184705B2Method and apparatus for motion compensated filtering of video signals

Publication Date: 2012.05.22 APTINA IMAGING CORP

AI summary

A method and apparatus for filtering video data. First and second frames of the video data are stored. Motion adapted spatio-temporal filter (MASTF) pixel values of the second frame are calculated using pixel values of the first and second frames. The second frame is compared to the first frame to estimate motion vectors (MVs) for the second frame. Pixel MV error level indicators for the second frame are determined using the pixel values of the first and second frames and the MVs of pixels in the second frame. Motion compensated temporal filter (MCTF) pixel values of the second frame are calculated using the pixel values of the first and second frames and the MVs of pixels in the second frame. For each pixel in the second frame, a filtered pixel value is calculated using its MASTF and MCTF pixel values and its pixel MV error level indicator.