Image Priority Ranking via Object Frequency Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional image ranking methods often misrank images by inadvertently prioritizing individuals of little importance due to positive facial expressions or favorable capture conditions, leading to irrelevant images being highly ranked.

Innovation Solution

An image management device calculates object priority based on the occurrence frequency of individuals within clusters, ranking images by the frequency of important persons, ensuring that images with frequently occurring important individuals are prioritized.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If images are ranked by evaluating facial expressions or capture conditions, then images with positive expressions are highly ranked, but irrelevant images featuring unimportant persons may be incorrectly prioritized

Engineering Contradiction:
Improveease of image selectionVSAvoidaccuracy of image priority ranking
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent segments the image ranking process into multiple independent evaluation dimensions: object priority (based on person importance and occurrence frequency), image quality priority (based on capture conditions and facial expressions), and composite image priority (combining both dimensions). This segmentation allows the system to evaluate different aspects separately and combine them, preventing any single dimension from causing incorrect rankings.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a new parameter dimension (object priority based on person identity and occurrence frequency) to complement the existing image quality parameters. By changing from a single-parameter ranking system to a multi-parameter system with different weightings, the patent resolves the contradiction between ease of selection and ranking accuracy.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If conventional ranking methods prioritize images with positive facial expressions, then image selection becomes easy, but the reliability of identifying important images decreases

Engineering Contradiction:
Improveefficiency of image searchVSAvoidprecision of image importance evaluation
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent divides the evaluation into separate modules: one for measuring image quality (facial expressions, capture conditions) and another for measuring object importance (person identity, occurrence frequency). Each module operates independently with its own precision metrics, and the results are combined to achieve both search efficiency and evaluation precision.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary evaluation layer that processes both image quality metrics and object priority metrics before producing the final ranking. This intermediary layer ensures that neither dimension dominates the other, maintaining both search efficiency and evaluation precision through balanced integration.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP2528034B1Image management device, image management method, program, recording medium, and integrated circuit
Publication Date: 2019.03.06 PANASONIC INTELLECTUAL PROPERTY CORP OF AMERICA
  • EP2528034B1 patent drawingFigure 1
  • EP2528034B1 patent drawingFigure 2
  • EP2528034B1 patent drawingFigure 3

AI summary

An image management device acquires an image group with an image acquisition unit, extracts objects and feature amounts from each image in the image group with an object detection unit, and sorts the objects into relevant clusters with an object sorting unit. Next, a similarity calculation unit calculates a similarity between the feature amounts of each object and each relevant cluster, a co-occurrence information generation unit finds co-occurrence information for each cluster, and then an accuracy calculation unit and an evaluation value calculation unit find an evaluation value for each object with respect to each cluster from the similarity and co-occurrence information. A n object priority evaluation unit evaluates the object priority of each object with the evaluation value, and an image priority evaluation unit evaluates the priority of each image from the object priority.