# US 2001/0025283 A1 · Apparatus and method for vector descriptor representation and multimedia data retrieval

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

- Number: US 2001/0025283 A1
- Document ID: US20010025283A1
- Status: Published application
- Kind: A1
- Application number: 09/808,627
- Filed: 2001-03-14
- Published: 2001-09-27
- Earliest claimed: 2000-03-18
- Assignees: Hyundai Electronics Ind. Co., Ltd.; Telecommunications Research Institute
- Inventors: Dong Gyu Sim; Hae Kwang Kim; Mun Churl Kim; Jin Woong Kim
- Web page: https://patentgrep.com/patent/US20010025283A1

## CPC

- G06F 16/40: of multimedia data, e.g. slideshows comprising image and additional audio data (retrieval of still image data G06F16/50; retrieval of audio data G06F16/60; retrieval of video data G06F16/70)
- G06F 16/583: using metadata automatically derived from the content
- G06F 18/2163: {Partitioning the feature space}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

The present invention relates to an apparatus and a method for vector descriptor representation and multimedia data retrieval, which can quantize a plurality of feature values described by a vector descriptor respectively, represent the quantized feature values in the form of bit or orthogonally transform the quantized bector feature values, and rearrange the feature values represented in the form of bit from the highest bit to the lowest bit or rearrange the transformed coefficient from low frequency to high frequency to represent the vector descriptor hierarchically. Moreover, when retrieving multimedia data, the present invention can code in variable length and store the rearranged feature values and the number of feature values which are input, inversely code only the feature values corresponding to the number of the feature values of the stored feature values, inversely arrange the inversely coded feature values to be restored to original feature values, inversely quantize the restored feature values, and compare the feature values restored by the inverse quantization with the feature values stored in a multimedia database to retrieve multimedia data.

## Claims

### Claim 1

A method for vector descriptor representation and multimedia data retrieval, the method comprising:
a quantization step of quantizing a plurality of feature values described by a vector descriptor respectively;
a bit representation step of representing each of the quantized feature values in the form of bit;
a bit rearrangement step of rearranging the feature values represented in the form of bit from the highest bit to the lowest bit and representing the vector descriptor hierarchically;
a variable-length coding step of coding in variable length and storing the rearranged feature values and the number of feature values which are input;
a variable-length inversely coding step of inversely coding only the feature values corresponding to the number of the feature values of the stored feature values;
a bit inverse arrangement step of inversely arranging the inversely coded feature values and restoring to original feature values;
an inverse quantization step of inversely quantizing the restored feature values; and
a comparison step of comparing the feature values restored by the inverse quantization with the feature values stored in a multimedia database and retrieving multimedia data.

### Claim 2

A method for vector descriptor representation and multimedia data retrieval, the method comprising:
an orthogonal transformation step of orthogonally transforming feature values described by a vector descriptor;
a feature value representation step of representing the transformed feature values from low frequency feature to high frequency feature;
a quantization step of quantizing the feature values represented in the feature value representation step;
a variable-length coding step of variable-length coding and storing the quantized feature values and the number of feature values which are input;
a variable-length inversely coding step of extracting the feature values corresponding to the number of the feature values of the stored feature values and inversely coding the extracted feature values;
an inverse quantization step of inversely quantizing the feature values inversely coded;
an inversely orthogonal transformation step of inversely and orthogonally transforming the inversely quantized feature values and restoring to original feature; and
a comparison step of comparing the restored feature values with feature values stored in a multimedia database and retrieving multimedia data.

### Claim 3 (depends on 2)

The method as claimed in claim 2, wherein the orthogonal transformation in the orthogonal transformation step uses DCT (Descrete Cosine Transform).

### Claim 4 (depends on 2)

The method as claimed in claim 2, wherein the orthogonal transformation in the orthogonal transformation step uses DST (Discrete Sine Transform).

### Claim 5 (depends on 2)

The method as claimed in claim 2, wherein the orthogonal transformation in the orthogonal transformation step uses DFT (Discrete Fourier Transform).

### Claim 6 (depends on 2)

The method as claimed in claim 2, wherein the orthogonal transformation in the orthogonal transformation step uses Haar

### Claim 7 (depends on 2)

The method as claimed in claim 2, wherein the orthogonal transformation in the orthogonal transformation step uses Wavelet.

### Claim 8 (depends on 2)

The method as claimed in claim 2, wherein the inversely orthogonal transformation in the inversely orthogonal transformation step uses inverse DCT.

### Claim 9 (depends on 2)

The method as claimed in claim 2, wherein the inversely orthogonal transformation in the inversely orthogonal transformation step uses inverse DST.

### Claim 10 (depends on 2)

The method as claimed in claim 2, wherein the inversely orthogonal transformation in the inversely orthogonal transformation step uses inverse DFT.

### Claim 11 (depends on 2)

The method as claimed in claim 2, wherein the inversely orthogonal transformation in the inversely orthogonal transformation step uses inverse Haar.

### Claim 12 (depends on 2)

The method as claimed in claim 2, wherein the inversely orthogonal transformation in the inversely orthogonal transformation step uses inverse Wavelet.

### Claim 13

An apparatus for vector descriptor representation and multimedia data retrieval, the apparatus comprising:
a quantization unit for quantizing a plurality of feature values described by a vector descriptor respectively;
a bit representing unit for representing each of the quantized feature values in the form of bit;
a bit rearranging unit for rearranging the feature values represented in the form of bit from the highest bit to the lowest bit and representing the vector descriptor hierarchically;
a variable-length coding unit for coding in variable length and storing the rearranged feature values and the number of feature values which are input;
a variable-length inversely coding unit for inversely coding only the feature values corresponding to the number of the feature values of the stored feature values;
a bit inverse arranging unit for inversely arranging the inversely coded feature values and restoring to original feature values;
an inverse quantization unit for inversely quantizing the restored feature values; and
a comparing unit for comparing the feature values restored by the inverse quantization with the feature values stored in a multimedia database and retrieving multimedia data.

### Claim 14

An apparatus for vector descriptor representation and multimedia data retrieval, the apparatus comprising:
an orthogonal transformation unit for orthogonally transforming feature values described by a vector descriptor;
a feature value representing unit for representing the transformed feature values from low frequency feature to high frequency feature;
a quantization unit for quantizing the feature values represented in the feature value representation step;
a variable-length coding unit for variable-length coding and storing the quantized feature values and the number of the feature values which are input;
a variable-length inversely coding unit for extracting the feature values corresponding to the number of the feature values of the stored feature values and inversely coding the extracted feature values;
an inverse quantization unit for inversely quantizing the feature values inversely coded;
an inversely orthogonal transformation unit for inversely and orthogonally transforming the inversely quantized feature values and restoring to original feature; and
a comparing unit for comparing the restored feature values with feature values stored in a multimedia database and retrieving multimedia data.

### Claim 15 (depends on 14)

The apparatus as claimed in claim 14, wherein the orthogonal transformation in the orthogonal transformation unit uses DCT (Descrete Cosine Transform).

### Claim 16 (depends on 14)

The apparatus as claimed in claim 14, wherein the orthogonal transformation in the orthogonal transformation unit uses DST (Discrete Sine Transform).

### Claim 17 (depends on 14)

The apparatus as claimed in claim 14, wherein the orthogonal transformation in the orthogonal transformation unit uses DFT (Discrete Fourier Transform).

### Claim 18 (depends on 14)

The apparatus as claimed in claim 14, wherein the orthogonal transformation in the orthogonal transformation unit uses Haar

### Claim 19 (depends on 14)

The apparatus as claimed in claim 14, wherein the orthogonal transformation in the orthogonal transformation unit uses Wavelet.

### Claim 20 (depends on 14)

The apparatus as claimed in claim 14, wherein the inversely orthogonal transformation in the inversely orthogonal transformation unit uses inverse DCT.

### Claim 21 (depends on 14)

The apparatus as claimed in claim 14, wherein the inversely orthogonal transformation in the inversely orthogonal transformation unit uses inverse DST.

### Claim 22 (depends on 14)

The apparatus as claimed in claim 14, wherein the inversely orthogonal transformation in the inversely orthogonal transformation unit uses inverse DFT.

### Claim 23 (depends on 14)

The apparatus as claimed in claim 14, wherein the inversely orthogonal transformation in the inversely orthogonal transformation unit uses inverse Haar.

### Claim 24 (depends on 14)

The apparatus as claimed in claim 14, wherein the inversely orthogonal transformation in the inversely orthogonal transformation unit uses inverse Wavelet.
