Browse Prior Art Database

Comparative Information Based Video and Image Object Search

IP.com Disclosure Number: IPCOM000240986D
Publication Date: 2015-Mar-16
Document File: 3 page(s) / 33K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for a user to perform a comparative information based search of video or image object context attributes.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 3

Comparative Information Based Video and Image Object Search

In many instances, people want to search video or images to compare more than one object. The comparison of objects in a video or image is largely a manual practice in which the user performs a visual comparison of the object attributes (e.g., size, color, shape, texture, motion, etc.) Alternatively, the video or image might have meta data or tags available that can be used for existing search methods; however, in many cases, the meta data or tags are not available or do not include the type of data needed for comparison.

A method is needed for performing a search query for comparison of more than one object based on the context attributes of a video or image.

The novel contribution is a method for a user to perform a comparison of information based on a search of video or image object context attributes. The method comprises an analysis of image of video content through image analytics, object recognition, and natural language processing to determine comparative attributes between two objects (e.g., size, proportion, speed, etc.), suitable for a media search whereby video and image objects are ranked and assigned confidence levels based on comparative attributes.

In a typical embodiment, the user selects a source image or video as a means of defining the search criteria for similar images or video found in the target environment (e.g., Internet, social media, folder on system, etc.). The software analyzes the source image using image analytics, object recognition, and natural language processing (NLP) to determine the key context and attributes, such as the following, to define search criteria:


 Location


 Object identification


 Relative size or proportion of objects in percentage (e.g., big vs. small, Object A bigger than object B, proportion of 70% vs. 30%, etc.)


 Texture of objects


 Color of objects


 Associated sound with objects for video (e.g., bird song, person speaking, thunder, etc.)


 Motion or speed of objects (e.g., high or low speed, moving, etc.)


 Subject context

The software performs a comparative analysis on the search results returned for images and video for specific objects of same type (e.g., comparing car with car for objects of same type). The system displays the comparative information search results with the video and image objects ranked based on informational attribute similarity confidence level.

This comparative search of image attributes offers a new aspect for image search that

1


Page 02 of 3

goes significantly beyond tagging and object analysis of a single object. It allows a user to not only request images or videos that contain two or more items, but also specify how to represent said objects in comparison to each other.

In the previous embodiment, a source image is used as input criteria. In an alternate...