Image Retrieval Indexing via Cross-Page Text Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image retrieval systems fail to accurately recall images when the query spans multiple web pages, as they treat each page as an independent unit and do not aggregate relevant text descriptions across pages, leading to missed results for queries requiring both A and B to be present on different pages.
Innovation Solution
The method involves identifying groups of images with identical content, aggregating related texts from all source webpages, and establishing an inverted index based on these text descriptions, allowing for cross-page image retrieval by treating the image as a basic unit and aggregating relevant source webpages for indexing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image retrieval is performed based on text information from a single webpage, then the retrieval process is simple and fast, but cross-page hit images cannot be recalled accurately
Solution Approach 1:
The patent merges text information from multiple webpages by aggregating texts associated with identical images across different sources. The system collects text information from multiple webpages, groups images by identity, and combines their associated texts to form comprehensive descriptions. This allows the inverted index to capture cross-page relationships, enabling accurate retrieval of images that appear on multiple pages with different textual contexts.
2Measurement precision
If text information is aggregated from multiple webpages for each image, then retrieval accuracy improves, but the complexity of establishing the inverted index increases
Solution Approach 1:
The patent segments the indexing process into distinct stages: first collecting text information from multiple webpages, then grouping images by identity, and finally establishing the inverted index based on aggregated text. This segmentation allows the system to handle complex multi-page relationships through a structured, step-by-step approach rather than attempting to process all information simultaneously, making the overall process more manageable despite the increased complexity.
3Ease of manufacture
If a page including the image is used as a basic unit, then the indexing process is straightforward, but image-related text cannot be determined accurately when the image appears on multiple pages
Solution Approach 1:
The patent inverts the traditional indexing approach by switching from using webpages as basic units to using images as basic units. Instead of indexing text by webpage, the system collects all texts associated with each image across multiple webpages, aggregates them, and then creates the inverted index. This inversion enables comprehensive text description for images that appear on multiple pages, capturing all relevant textual contexts rather than limiting to a single page.
Data Source
AI summary
Embodiments of the present disclosure disclose a method, apparatus, server and storage medium for image retrieval. The method includes: identifying a plurality of groups of images having identical contents from images on all webpages; aggregating, for each image group, image-related texts on all source webpages of each image to obtain text descriptions of each image group; establishing an inverted index for each image in the image groups based on the text descriptions of the image group, the inverted index at least including, for each text description, source webpages corresponding to all text descriptions of the image group of the text description; and performing image retrieval based on an inputted query and the inverted index.