Browse Prior Art Database

Searching Images Based On Refining Desired Image Characteristics Disclosure Number: IPCOM000213425D
Publication Date: 2011-Dec-14
Document File: 3 page(s) / 38K

Publishing Venue

The Prior Art Database


Disclosed is a method to refine and improve an image search through an interface which provides users a pop-up rating system through which they can indicate which elements of the results they prefer.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 3

Searching Images Based On Refining Desired Image Characteristics

Image searching is becoming common, but locating the proper results can still be difficult if the user is not sure of what they need to find. Currently, users are limited to a few methods, the most common of which uses keywords for searching. A problem with

using keywords for searching images is if a user is not quite sure what qualities the image needs to have, it is nearly impossible to describe the desired search criteria. This method also relies on searching text surrounding the image versus qualities in the image itself.

A more recent development is providing an image as a search query and getting results

back. This is acceptable if the user is looking to identify an image that the user already has, but it is not useful for refining searches when looking for matches. This difficulty in locating desired images of a type can be very frustrating.

Somewhat relevant technologies that are in current use search include:

• General image search or taking an image and searching for similar ones, but doesn't mention refining search through specific characteristics. [1]

• Searching images based on sub-sections of an existing image, like selecting a region of a photo to identify a house, then selecting subsections of that to identify other elements in the photo, rather than qualities of the photo itself. [2]

• A method that goes with the thumbs-up or thumbs-down and keyword search, but does not really extend to the customization angle beyond the simple +/- . [3] The disclosed invention exposes the qualities of a selected image and the user can check or uncheck elements (e.g., colorful, hard angles, unsaturated, soft, etc) to refine the search instead of constantly drilling.

This invention relates to refining and improving an image search using an interface to allow users to define what elements of results they prefer through a rating system, thus gathering more similar results and refining the search. The search is started by a user providing some seed to reference. This could come from a standard keyword, or providing a sample image, or any number of ways. Once the user is presented with the initial set of results, the refinement can begin.

On the returned results, each one can be given a simple thumbs-up or thumbs down, selected through a pop-up tool, refreshing the results either immediately or after a user action or time period. (Figure 1) The refreshed results are based on those for which the user has shown a preference, which can be generated by computationally determining

features of the image, such as texture, contrast, types of shapes...