Image Trimming Control Using ML-Based Quality Evaluation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image processing technologies struggle to determine an appropriate trimming method for images efficiently, particularly in determining the extent to which to emphasize the main subject or maintain the composition, without user intervention.

Innovation Solution

An image processing device utilizes a machine learning model trained on user feedback to evaluate image quality and determine trimming methods based on user attributes, accessory information, and image specifications, applying different trimming techniques based on evaluation thresholds.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If automatic trimming is performed without user intervention, then productivity is improved, but manufacturing precision deteriorates

Engineering Contradiction:

Improveautomatic trimming efficiencyVSAvoidtrimming accuracy

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The system uses machine learning models trained on user feedback data to automatically determine trimming methods. The model learns from user preferences and corrections to accurately predict appropriate trimming parameters without requiring direct user intervention for each image, thus maintaining both automation and precision.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The trimming system performs self-service by automatically analyzing images and applying appropriate trimming methods based on learned patterns from training data. The system serves itself by using accumulated user feedback to improve its own decision-making capability, eliminating the need for continuous manual input while maintaining high accuracy.

Inventive Principle:

Principle #25Self-service

2Ease of operation

If simple trimming rules are used, then ease of operation is improved, but manufacturing precision deteriorates

Engineering Contradiction:

Improveautomatic trimming simplicityVSAvoidtrimming accuracy

Core Design Contradiction:

Ease of operationVSManufacturing precision

Solution Approach 1:

The system changes parameters dynamically based on image characteristics and user preferences stored in the training data. Instead of using fixed simple rules, the model adjusts trimming parameters such as crop boundaries, scaling factors, and composition ratios based on the specific image content and learned user preferences, achieving both simplicity and precision.

Inventive Principle:

Principle #35Parameter changes

3Adaptability or versatility

If user-specific trimming methods are implemented, then adaptability is improved, but device complexity increases

Engineering Contradiction:

Improveuser preference adaptationVSAvoidsystem complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

User preferences and trimming patterns are collected and processed in advance during a training phase. The system performs preliminary action by gathering user feedback data and training the machine learning model before actual trimming operations. This pre-processing of user preferences simplifies the runtime complexity while maintaining high adaptability to individual user needs.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20250342596A1Image processing device, operation method of image processing device, and operation program of image processing device

Publication Date: 2025.11.06 FUJIFILM CORP

AI summary

An image processing device includes: a processor, in which the processor is configured to: obtain an evaluation value for a quality of an image; and determine a trimming method of a trimming target image, which is any one of the image or a related image of the image, based on the evaluation value.