Unlock instant, AI-driven research and patent intelligence for your innovation.

Retrieval method based on mean value removal product quantization, storage medium and terminal equipment

A de-average and product technology, applied in the retrieval method, storage medium and terminal equipment based on de-average and product quantization, can solve the problem of low retrieval accuracy, and achieve the effect of improving accuracy and robustness

Pending Publication Date: 2020-03-17
PENG CHENG LAB +1
View PDF1 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by the present invention is to provide a retrieval method, storage medium, and terminal device based on the quantization of the product of the mean to solve the problem of retrieval accuracy when the variance of the sample mean value in the data set is large. low problem

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Retrieval method based on mean value removal product quantization, storage medium and terminal equipment
  • Retrieval method based on mean value removal product quantization, storage medium and terminal equipment
  • Retrieval method based on mean value removal product quantization, storage medium and terminal equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] This embodiment provides a retrieval method based on de-mean product quantization, such as figure 1 As shown, the method includes:

[0050] S10. Obtain a target retrieval sample, and calculate a residual vector of the target retrieval sample.

[0051]Specifically, the target retrieval sample is a request sample, several retrieval samples are pre-stored in the preset database, and the retrieval method is to retrieve a preset number of retrieval samples corresponding to the target retrieval sample from among the several retrieval samples. That is to say, the preset database is a search sample database corresponding to the target search sample, and a preset number of search samples corresponding to the target search sample need to be retrieved in the preset database. In this embodiment, the target retrieval samples are stored in the form of vectors, each of the several retrieval samples is stored in the form of vectors, and the vector dimension corresponding to each retri...

Embodiment 2

[0088] This embodiment provides a retrieval method based on de-mean product quantization, the method comprising:

[0089] M10. Obtain a target retrieval sample, and calculate a residual vector of the target retrieval sample.

[0090] Specifically, the target retrieval sample is a request sample, several retrieval samples are pre-stored in the preset database, and the retrieval method is to retrieve a preset number of retrieval samples corresponding to the target retrieval sample from among the several retrieval samples. That is to say, the preset database is a search sample database corresponding to the target search sample, and a preset number of search samples corresponding to the target search sample need to be retrieved in the preset database. In this embodiment, the target retrieval samples are stored in the form of vectors, each of the several retrieval samples is stored in the form of vectors, and the vector dimension corresponding to each retrieval sample is the same a...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a retrieval method based on mean value removal product quantification, a storage medium and terminal equipment, and the method comprises the steps: calculating the mean value of a target retrieval sample, and searching a target quantification mean value corresponding to the target retrieval sample in a preset database according to the mean value; and calculating a residualvector corresponding to the target retrieval sample according to the target quantization mean value, and retrieving a preset number of retrieval samples in a preset database according to the residualvector and the target quantization mean value. According to the invention, the target quantification mean value corresponding to the target retrieval sample is searched in the preset database; the residual vector of the target retrieval sample is determined according to the quantization mean value, and the retrieval sample corresponding to the target retrieval sample is determined through the residual vector and the target quantization mean value, so that the influence of sample mean value distribution on quantization through the residual vector is reduced, and the retrieval accuracy and robustness are improved.

Description

technical field [0001] The present invention relates to the technical field of retrieval, in particular to a retrieval method, storage medium and terminal equipment based on de-mean product quantization. Background technique [0002] In recent years, quantization methods have achieved remarkable results in the field of information retrieval, aiming to quantize information vectors into a small number of codewords, and only need to compare the similarity between the target vector or its corresponding codewords and the database codewords in the retrieval process. The vector quantization is to directly quantize the information vectors to generate codewords with the same dimensions as the information vectors. Each information vector corresponds to a unique codeword. However, when the number of information vectors is huge, the required codebook (set of codewords ) is also very large. Therefore, for large data sets, vector quantization still has the problem of high storage complex...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/2453
CPCG06F16/2453Y02D10/00
Inventor 夏树涛阳佳城陈斌戴涛李清汪漪
Owner PENG CHENG LAB