Image Search Using Qualitative and Quantitative Tag Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing applications struggle to effectively search and organize large datasets of images due to the inability to distinguish between similar object tags, leading to inefficient retrieval of desired images.
Innovation Solution
An image processing apparatus and method that evaluates image data to generate both qualitative and quantitative tags, allowing for the identification of similar images based on predetermined frequency and quality metrics, thereby improving search efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If common object tags are used for image search, then the search process is simple, but the search precision deteriorates due to large number of images with same tags
Solution Approach 1:
The patent segments image tags into multiple categories: common object tags (e.g., 'cat') and detailed attribute tags (e.g., 'orange tabby cat'). This segmentation allows users to search at different levels of detail, improving precision while maintaining operational simplicity through hierarchical tag structure.
Solution Approach 2:
The patent adds a new dimension to image search by introducing detailed attribute tags that describe specific characteristics of objects in images. This transforms the search space from simple object identification to multi-dimensional attribute-based search, enabling precise differentiation among images with common objects.
2Measurement precision
If detailed attribute tags are added to images, then search precision is improved, but the device complexity increases due to multiple tag types
Solution Approach 1:
The patent implements a universal tagging system where a single image can have multiple types of tags (object tags, attribute tags, quality tags). This multi-functional tag system handles diverse search requirements through a unified structure, reducing the need for separate processing systems for different tag types.
Solution Approach 2:
The patent applies preliminary action by automatically generating detailed attribute tags and quality assessments during image ingestion, rather than requiring manual tagging or complex real-time analysis. This pre-processing reduces the complexity of search operations by having all tag data ready in advance.
3Measurement precision
If image evaluation is performed to generate quality tags, then image quality assessment is improved, but the processing time increases
Solution Approach 1:
The patent performs image evaluation and quality tag generation as a preliminary action during image ingestion and storage. By completing the evaluation process before the search operation, the system avoids adding processing time to user search queries while maintaining high quality assessment precision.
Solution Approach 2:
The system implements self-service by automatically evaluating image quality and generating appropriate tags without requiring manual intervention or complex real-time processing during search operations. The pre-computed quality metrics serve the search function directly, reducing processing time.
Data Source
AI summary
An image processing method and apparatus is provided for obtaining an image captured by an image capturing apparatus; determining characteristic data of the captured image based on inputting the obtained image into an image analysis model trained with images including a predetermined type of characteristic data; and identifying, based on the determination performed by the image analysis model, metadata including a numeric value regarding the predetermined type of the captured image.