Image classification method based on kernel principal component analysis network

A technology of kernel principal component analysis and classification methods, applied in the field of digital images, can solve the problems of very sensitive training data noise, loss of nonlinear energy in high dimensions, lack of normalization of functions, etc.

Active Publication Date: 2015-04-29
SOUTHEAST UNIV
View PDF2 Cites 17 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

If γ is overestimated, the exponent will be nearly linear, and high-dimensional projections lose their nonlinear energy
Conversely, if γ is estima

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 classification method based on kernel principal component analysis network
  • Image classification method based on kernel principal component analysis network
  • Image classification method based on kernel principal component analysis network

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0170] The technical solution of the present invention will be described in detail below with reference to the drawings and embodiments.

[0171] Such as Figure 1-4 As shown, the image classification method based on the nuclear principal component analysis network in the present invention constructs a new image feature extraction structure by cascading two-layer nuclear principal component analysis filters, which is called the nuclear principal component analysis network. The nuclear principal component analysis network was tested, specifically including the following steps.

[0172] Step 1: Establish the first layer of the nuclear principal component analysis network

[0173] Step 1.1: Randomly select N from N image databases of size m×n 1 Frame as the training image database; use a size k 1 ×k 2 The slider of traverses each training image in the training image database Each pixel of which Is the set of real numbers, k 1 And k 2 Are all odd and 0 1 ≤m, 0 2 ≤n, each image has a t...

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 classification method based on a kernel principal component analysis network. The image classification method comprises the following steps that (1) a training image is input and pre-processed to obtain a local characteristic matrix of the training image; (2) the two-layer kernel principal component analysis network is built to obtain a main characteristic vector of the training image; (3) a classifier is trained by using the obtained main characteristic vector; for verifying the correctness of the classification, a kernel principal component analysis testing network is built to test a test image. According to the image classification method, through building the two-layer kernel principal component analysis network, the nonlinear characteristics of the image can be obtained, therefore, the description of the characteristics of the image is more precise, the classification is more accurate, and the accuracy is higher for image classification problems.

Description

technical field [0001] The invention relates to the field of digital images, in particular to an image classification method based on kernel principal component analysis network. Background technique [0002] Before classifying an image, it is often necessary to reduce the dimensionality of the image matrix. There are two commonly used dimensionality reduction methods: 1. Principal component analysis; 2. Kernel principal component analysis. The former is mainly for linearly separable data, while the latter can handle linearly inseparable data. The classification accuracy of both is not very high. [0003] 1. Principal Component Analysis [0004] Principal component analysis is the most commonly used linear dimensionality reduction method. Its goal is to map high-dimensional data into a low-dimensional space for expression through linear mapping, and it is expected that the variance of the data on the projected dimension will be the largest. Fewer data dimensions retain m...

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): G06K9/62
Inventor 吴丹伍家松姜龙玉杨淳沨达臻舒华忠
Owner SOUTHEAST UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap