# US 5,708,805 A · Image retrieving apparatus using natural language

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 5,708,805 A
- Document ID: US5708805A
- Status: Granted
- Kind: A
- Application number: 08/135,294
- Filed: 1993-10-12
- Granted: 1998-01-13
- Earliest claimed: 1992-10-09
- Assignees: Matsushita Electric Industrial Co., Ltd.
- Inventors: Shusaku Okamoto; Masahiro Hamada
- Web page: https://patentgrep.com/patent/US5708805A

## CPC

- G06F 16/58: Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F 16/583: using metadata automatically derived from the content
- Y10S 707/99933: Query processing, i.e. searching
- Y10S 707/99945: Object-oriented database structure processing
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y10S: TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS

## Abstract

An image retrieving apparatus including a language analyzer, an object determiner, a rule for defining an undefined object, an image selector, a display, a word dictionary, and an image database. The language analyzer analyzes an input sentence and extracts objects of the input sentence. The object determiner determines whether the object data expresses a background or a part using a rule for defining an undefined object. Images of background and parts are separately retrieved.

## Claims

### Claim 1

An image retrieving apparatus comprising:
a word dictionary that stores information for analyzing a sentence;
an image database that stores image components, each image component comprising a pixel data and an attribute, said pixel data having characteristics, said attribute indicating quantitatively said characteristics of said pixel data;
object controlling means for controlling object data for an object, said object data comprising a) a name for said object that expresses an entity, b) a common knowledge concerning said object, and c) a pointer that points to said image components in said image database that express said entity of said object;
sentence input means for inputting a sentence;
a language analyzing means for a) analyzing an input sentence using said word dictionary, b) extracting objects of said input sentence, and c) developing an image expression containing said object data from said object controlling means with respect to said extracted objects; and
displaying means for displaying said pixel data of said image components pointed to by a plurality of said object data of said image expression.

### Claim 2 (depends on 1)

An image retrieving apparatus according to claim 1, wherein said attribute of each of said image components has at least a name corresponding to said entity expressed by said pixel data of said image component and wherein said attribute for each entity is different.

### Claim 3 (depends on 1)

An image retrieving apparatus according to claim 1, wherein at least one of said object data has a conceptually lower-level object and a second pointer corresponding to said lower-level object.

### Claim 4 (depends on 1)

An image retrieving apparatus according to claim 1, wherein the plurality of said object data of said image expression is all of said object data of said image expression.

### Claim 5 (depends on 1)

An image retrieving apparatus according to claim 1, wherein the plurality of said object data of said image expression is a maximum number of said object data of said image expression.

### Claim 6

An image retrieving apparatus comprising:
a word dictionary that stores information for analyzing a sentence;
an image database that stores image components, each image component comprising a pixel data and an attribute, said pixel data having characteristics, said attribute indicating quantitatively said characteristics of said pixel data;
object controlling means for controlling object data for an object, said object data comprising a) a name for said object that expresses an entity, b) a common knowledge concerning said object, and c) a pointer that points to said image components in said image database that express said entity of said object;
sentence input means for inputting a sentence;
a language analyzing means for a) analyzing an input sentence using said word dictionary, b) extracting objects of said input sentence, and c) developing an image expression containing said object data from said object controlling means with respect to said extracted objects;
object discriminating means, using said common knowledge from said object controlling means, for discriminating whether each one of said extracted objects is a background or a part;
image selecting means for a) receiving a plurality of said object data from said image expression, b) determining if said plurality of said object data express said background, c) extracting said pixel data of said image components pointed to by said pointer of said object data which express said background and d) selecting only pixel data that are pointed to by said object data; and
displaying means displaying said selected pixel data of said image components pointed to by said pointer of said object data of said image expression.

### Claim 7 (depends on 6)

An image retrieving apparatus according to claim 6, wherein said object discriminating means has a rule for determining whether an undefined object is a background or a part.

### Claim 8 (depends on 7)

An image retrieving apparatus according to claim 7, wherein said image components further include discrimination information for identifying whether said object which expresses said pixel data is said background or said part, and
said object discriminating means, further using said discrimination information, for discriminating whether said object data is said background or said part.