Foreground-Aware Image Redimensioning with ML Background Extension

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional digital image editing systems are limited to editing content within the digital image, leading to visual inaccuracies, computational inefficiencies, and increased power consumption, and cannot expand or reposition content beyond the image's dimensions.

Innovation Solution

A re-dimension system that segments foreground objects from the background using a machine learning model, adjusts the background dimensions, fills holes or extends the background as needed, positions the foreground object optimally, and harmonizes the object with the background to create a cohesive re-dimensioned image.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If conventional digital image editing systems are used to edit content within the digital image, then the editing process is simple and straightforward, but visual inaccuracies occur and important content is lost when re-dimensioning is required

Engineering Contradiction:

Improvevisual accuracyVSAvoidsystem complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system segments the digital image into foreground objects and background using machine learning models. This segmentation allows independent processing of foreground and background, enabling accurate re-dimensioning by filling background holes and extending foreground objects without losing important content, thereby resolving the visual accuracy issue while maintaining manageable system complexity through modular architecture.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces machine learning models as intermediary components between the input image and the re-dimensioning process. These models (segmentation models, generation models) act as mediators that intelligently separate and reconstruct image components, achieving high visual accuracy without requiring complex manual editing operations.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Loss of information

If the digital image dimension is updated using conventional editing methods, then the process is computationally efficient, but the foreground object may be cropped or lost

Engineering Contradiction:

Improvecontent retentionVSAvoidprocessing efficiency

Core Design Contradiction:

Loss of informationVSProductivity

Solution Approach 1:

The system performs preliminary segmentation of the foreground object from the background before re-dimensioning. By pre-identifying and separating the foreground object, the system can preserve it during dimension updates, then reposition and re-integrate it into the re-dimensioned background, ensuring no content is lost while maintaining processing efficiency through automated workflows.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent changes the parameter of image dimension while using machine learning models to predict and generate missing foreground object portions. This allows the system to adapt to new dimension parameters without losing important content, balancing information retention with processing efficiency through intelligent parameter adaptation.

Inventive Principle:

Principle #35Parameter changes

3Area of stationary object

If the digital image is re-dimensioned to expand beyond original dimensions, then more content can be displayed, but computational inefficiencies and increased power consumption occur

Engineering Contradiction:

Improveimage areaVSAvoidpower consumption

Core Design Contradiction:

Area of stationary objectVSUse of energy by moving object

Solution Approach 1:

The system uses self-service machine learning models that automatically perform segmentation, background hole filling, and foreground object extension without requiring manual intervention. This automation achieves efficient image area expansion while optimizing power consumption by using intelligent algorithms that adapt to the specific image content, processing only what is necessary for the re-dimensioning task.

Inventive Principle:

Principle #25Self-service

4Adaptability or versatility

If conventional image editing is used, then the system is easy to operate, but it cannot reposition or expand content beyond the image's original dimensions

Engineering Contradiction:

Improvere-dimensioning capabilityVSAvoidoperational simplicity

Core Design Contradiction:

Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent implements a universal re-dimensioning system that handles multiple operations (segmentation, background filling, foreground extension, repositioning) through a single integrated machine learning-based workflow. This multi-functional approach provides versatile re-dimensioning capabilities while maintaining ease of operation by automating complex tasks that would otherwise require multiple manual editing steps.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250225609A1Re-dimensioning images based on foreground objects

Publication Date: 2025.07.10 ADOBE INC

AI summary

In implementation of techniques for re-dimensioning images based on foreground objects, a computing device implements a re-dimension system to receive a digital image and an input specifying an update to a dimension of the digital image. The re-dimension system then generates, using the machine learning model, a re-dimensioned background by changing the background based on the update to the dimension specified by the input. Using the machine learning model, the re-dimension system generates a re-dimensioned digital image by positioning the foreground object over the re-dimensioned background. The re-dimension system then displays the re-dimensioned digital image in a user interface.