Neural Network Motion Deblurring via Foreground-Background Segmentation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing computer vision systems struggle to accurately remove motion blur from images, particularly when objects in the foreground and background experience different types of motion blur, leading to degraded image quality.

Innovation Solution

A neural network architecture comprising a human-aware attention model, encoder network, and multi-branch decoder network is trained to generate masks and feature information for foreground and background portions of images, allowing for separate deblurring operations to reconstruct enhanced images with reduced motion blur.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If a single deblurring operation is applied to the entire image, then the processing is simple and fast, but the foreground and background portions cannot be differentiated leading to poor deblurring quality

Engineering Contradiction:

Improvedeblurring qualityVSAvoidprocessing complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The image is segmented into foreground and background portions using a human-aware attention model that generates separate masks for each region. This segmentation allows the system to apply different deblurring operations to different parts of the image, improving overall deblurring quality while managing complexity through specialized processing for each segment.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different deblurring operations are applied to different regions of the image based on their specific characteristics. The foreground portion receives one type of deblurring treatment while the background portion receives another, allowing each region to be optimized for its specific motion blur characteristics rather than applying a uniform approach.

Inventive Principle:

Principle #3Local quality

2Manufacturing precision

If separate deblurring operations are applied to foreground and background portions, then the deblurring quality is improved, but the processing time and computational complexity increase

Engineering Contradiction:

Improveimage reconstruction accuracyVSAvoidprocessing time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The human-aware attention model performs preliminary classification of image regions into foreground and background portions before the actual deblurring process. By pre-segmenting the image and identifying which regions require which type of deblurring treatment, the system avoids unnecessary computational overhead during the main processing stage and can optimize the processing pipeline accordingly.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10593021B1Motion deblurring using neural network architectures

Publication Date: 2020.03.17 INCEPTION AI IP LTD

AI summary

This disclosure relates to improved techniques for performing computer vision functions including motion deblurring functions. The techniques described herein utilize a neural network architecture to perform these functions. The neural network architecture can include a human-aware attention model that is able to distinguish between foreground human objects and background portions of degraded images affected by motion blur. The neural network architecture further includes an encoder-decoder network that separately performs motion deblurring functions on foreground and background portions of degraded images, and reconstructs enhanced images corresponding to the degraded images.