Content-based image search

Abstract

Image descriptor identifiers are used for content-based search. A plurality of descriptors is determined for an image. The descriptors represent the content of the image at respective interest points identified in the image. The descriptors are mapped to respective descriptor identifiers. The image can thus be represented as a set of descriptor identifiers. A search is performed on an index using the descriptor identifiers as search elements. A method for efficiently searching the inverted index is also provided. Candidate images that include at least a predetermined number of descriptor identifiers that match those of the image are identified. The candidate images are ranked and at least a portion thereof are presented as content-based search results.

Description

Background technique [0001] Many methods are known in the art for searching and retrieving information, such as over the Internet using search engines. Such methods typically employ text-based searches. Text-based searches employ a search query that includes one or more textual elements such as words or phrases. The text elements are compared to an index or other data structure to identify web pages, documents, images, etc. that include matching or semantically similar text content, metadata, filenames, or other text representations. [0002] Known text-based search methods work relatively well for text-based documents, but are difficult to apply to image files and data. In order for an image file to be searched by a text-based query, the image file must be associated with one or more textual elements, such as a title, filename, or other metadata or tags. The search engines and algorithms employed for text-based searches are unable to search image files based on the content...

AI Technical Summary

Problems solved by technology

But these methods are rather cumbersome and may require extensive input from the user to characterize the image before a search can be performed

Furthermore, such methods are inefficient and do not scale well to large-scale applications, where for example billions of images must be quickly searched in order to identify and provide search result images to users

Images