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

Image feature extraction method based on complex values

An image feature extraction and complex value technology, which is applied in the direction of instruments, biological neural network models, character and pattern recognition, etc., can solve the problem of image feature extraction is not perfect, so as to improve the representation effect, good robustness and representation, Solve the effect that image features are not accurate enough

Pending Publication Date: 2020-11-13
UNIV OF ELECTRONIC SCI & TECH OF CHINA
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide an image feature extraction method based on complex values, which can make image features more expressive and solve the problem that the current image feature extraction is not perfect

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
  • Image feature extraction method based on complex values
  • Image feature extraction method based on complex values
  • Image feature extraction method based on complex values

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0086] The image feature extraction method based on complex value of the present embodiment, comprises the steps:

[0087] S11: Complex BN layer (Complex BN).

[0088] In the structure of real numbers, the calculation formula of BN is:

[0089]

[0090] In the formula, β=E[x]. is the normalization of the current sample x in the whole batch X, expressed as In this embodiment, the covariance matrix V is used to calculate, and the formula is as follows:

[0091]

[0092]

[0093] where R(x) and I(x) denote the real and imaginary parts of the eigenvector x, respectively. In order to simplify the calculation process, the BN algorithm is used for the real part and the imaginary part respectively. The expression is as follows:

[0094]

[0095] The improved ComplexBN calculation method has obtained a certain performance improvement.

[0096] S12: CRelu layer.

[0097] The proposed CRelu layer complex-valued activation method is the best activation method. Spli...

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 an image feature extraction method based on a complex value, and belongs to the field of image feature extraction. In order to enable image features to be more expressive and solve the problem that the current image feature extraction is incomplete, the method comprises the following steps: constructing a neural network complex value layer based on a plurality of images; constructing a plurality of modules for feature extraction by utilizing the replication layer; and combining the plurality of modules, and performing image feature extraction by using the combined modules. On the basis of the existing neural network structure, the image feature representation effect is greatly improved by introducing a plurality of data expressions.

Description

technical field [0001] The invention relates to the field of image feature extraction, in particular to an image feature extraction method based on complex values. Background technique [0002] Nowadays, with the continuous innovation of information technology, computer vision is a very important field, and images are the basis of computer vision. In order to better express images, excellent image feature extraction methods are the most important. [0003] The complex-value network is a deep learning network based on complex numbers. At present, the great achievement of convolutional neural network (CNN) in feature and metric learning has attracted many researchers. However, the vast majority of deep network architectures are based on real-valued representations. Due to the lack of effective models and suitable distances for complex-valued vectors, research on complex-valued networks has received little attention. Research has shown that complex vectors have richer repres...

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
IPC IPC(8): G06K9/46G06K9/62G06N3/04
CPCG06V10/40G06N3/048G06N3/045G06F18/22G06F18/213G06F18/2411
Inventor 赵太银秦科田玲罗光春魏文轩刘江麟
Owner UNIV OF ELECTRONIC SCI & TECH OF CHINA