Image Search System Using Region-Specific Tag Association

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image search systems are inefficient and error-prone, as they require users to manually identify and filter search results for specific objects and locations within images, leading to frustration and resource wastage.

Innovation Solution

An image search method and apparatus that receives a target object and area input, analyzes tag information to identify and display images where the object is situated at a specific location or geographic area, facilitating precise searches by associating target area information with image metadata.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If users manually identify and filter search results for specific objects and locations, then search flexibility is maintained, but search efficiency and accuracy deteriorate

Engineering Contradiction:
Improvesearch efficiencyVSAvoidsearch accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary analysis of images to automatically identify objects, people, and locations, and pre-associates tag information with specific regions within images. This preliminary action enables users to directly search using these pre-extracted features without manual identification, thereby improving both search efficiency and accuracy simultaneously

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary tagging system that bridges between raw image data and user search queries. The tag information acts as a mediator, containing structured data about objects, people, and locations with their spatial relationships, allowing users to search efficiently while maintaining precision through the structured intermediary layer

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If basic tag information is used for image search, then search simplicity is maintained, but search precision for location-specific queries deteriorates

Engineering Contradiction:
Improvesearch precisionVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments tag information into multiple hierarchical levels: basic tags for overall image content, and region-specific tags for objects, people, and locations within specific areas of images. This segmentation allows the system to provide detailed location-specific search precision while keeping the basic interface simple for users

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent adds a spatial dimension to traditional tag information by associating tags with specific regions, coordinates, and geographic locations within images. This dimensional enhancement enables precise location-based searching without requiring complex user interactions, as the spatial information is automatically processed by the system

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Reliability

If multiple search criteria are applied separately, then search thoroughness is improved, but search process complexity and error probability increase

Engineering Contradiction:
Improvesearch reliabilityVSAvoidsearch process complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent merges multiple search criteria (object identification, location detection, geographic positioning) into a unified tag information structure. This consolidation allows users to perform comprehensive searches using a single integrated query process, improving search reliability while reducing process complexity and error probability

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The tag information system is designed with multi-functionality, serving as a universal data structure that simultaneously supports object-based search, location-based search, and combined searches. This universal approach eliminates the need for separate search processes for different criteria, thereby improving reliability while simplifying the user interface and workflow

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP2733629B1System for associating tag information with images supporting image feature search
Publication Date: 2019.04.17 SAMSUNG ELECTRONICS CO LTD
  • EP2733629B1 patent drawingFigure 1A~1B
  • EP2733629B1 patent drawingFigure 1C~1E
  • EP2733629B1 patent drawingFigure 2

AI summary

An image search method and apparatus. The method comprises receiving search criteria comprising a name of a target object and a target area within an image, analysing tag information of a plurality of images to identify candidate images including the target object at least partially within the target area and displaying data identifying the candidate images as search results.