Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Multi-label natural scene classification method based on spatial pyramid and sparse coding

A spatial pyramid and sparse coding technology, applied in the field of multi-label classification of natural scenes based on spatial pyramid sparse coding, can solve the problem that it is difficult to obtain complete and correct classification of images, and achieve high accuracy, good robustness, and robust classification good sex effect

Active Publication Date: 2015-11-18
XIDIAN UNIV
View PDF2 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The above methods only use color information as the feature vector in the multi-label classification problem, so it is difficult to obtain a complete and correct classification of the image when the image is translated, rotated, brightness and scale change

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
  • Multi-label natural scene classification method based on spatial pyramid and sparse coding
  • Multi-label natural scene classification method based on spatial pyramid and sparse coding
  • Multi-label natural scene classification method based on spatial pyramid and sparse coding

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0030] The invention is a multi-label classification method for natural scenes based on spatial pyramid sparse coding. refer to figure 1 , the specific implementation steps of the present invention are as follows:

[0031] Step 1, establish a multi-label category library for natural scene images, and use images as input data.

[0032] The natural scene image multi-label category library original is used as the image library used in the experiment of the present invention. The image library contains 2000 pieces of natural scene images, and all possible concepts are marked as desert, mountain, ocean, sunset and tree, and each image is manually marked A collection of concept tags. Images with two or more concept labels (such as "ocean+sunset") account for about 22% of the image library, and each image corresponds to an average of 1.24±0.44 concept labels.

[0033] Step 2, for each image in the multi-label category library of natural scene images, extract the scale-invariant SI...

Embodiment 2

[0047] Take k pixels as the step size to uniformly sample, and extract the d-dimensional scale-invariant feature SIFT of the 16×16 pixel image block around each sampling point. In this example, use 6 pixels as the step size to sample uniformly, and extract each sampling point The d-dimensional scale-invariant feature SIFT of the surrounding 16×16 pixel size image block,

[0048] The natural scene multi-label classification method based on spatial pyramid sparse coding is the same as embodiment 1, wherein the d-dimensional scale-invariant feature SIFT of extracting the 16 × 16 pixel size image block around each sampling point described in step 2 is carried out as follows:

[0049] (2.1) Gaussian filtering is performed on an image block with a size of 16×16 pixels, wherein the parameters of the Gaussian filtering are: the mean value is 0, the variance is 1, and the size is 5×5 pixels;

[0050] (2.2) Calculate the gradient modulus and gradient direction of each pixel in the Gauss...

Embodiment 3

[0053] The natural scene multi-label classification method based on spatial pyramid sparse coding is the same as embodiment 1-2, wherein the SIFT feature matrix to all images described in step 3, randomly selects M feature vectors therefrom to form a new feature matrix Y, which is passed through K-singular value decomposition method K-SVD training dictionary D, proceed as follows:

[0054] (3.1) Randomly select M eigenvectors from the SIFT eigenmatrices of all images to form a new eigenmatrix Y, the size of which is 128×M, wherein M generally takes values ​​of 100000, 200000, 1000000, and M=200000 in this embodiment , the invention selects M feature vectors from the feature matrix for dictionary training, which reduces the amount of training calculations.

[0055] (3.2) randomly select the B column eigenvectors in the matrix Y to initialize the dictionary D, the number of atoms of the dictionary B=1024 in the present embodiment, the size of the dictionary D is 128 * 1024, the ...

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 multi-label natural scene classification method based on spatial pyramid and sparse coding, and mainly aims at solving the problems that a present classification method cannot completely describe a natural scene and the classification accuracy is relatively low. The method comprises the steps that a multi-label class library of a natural scene is established; the scale invariant feature (SIFT) of the class library is extracted to generate a sparse dictionary D; the sparse dictionary is used to carry out dictionary mapping on the image, and the spatial pyramid and sparse coding are used to generate a multi-scale sparse vector; and a classification result of a multi-class support vector machine is used to correct and order classification results of a support vector machine, and further to obtain the final classification result of the natural scene image. The multi-scale feature, sparse coding and multi-scale classification method is used, local information of the image is extracted, characteristic information of the is enriched, the natural scene is described more comprehensively, the classification precision and robustness of the natural scene are improved, and the method can be used to match, classify and identify the natural scenes.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a natural scene classification method for image translation, rotation, brightness and scale change, specifically a multi-label classification method for natural scenes based on spatial pyramid sparse coding, which can be used for natural scene matching of images, classification and identification. Background technique [0002] In the past decade, natural scene image classification has become a very important technical problem in the field of image processing. Natural scene image classification has a wide range of applications, such as target recognition and detection, intelligent vehicle or robot navigation and other fields. Due to the large differences in the categories of natural scene images, differences in lighting conditions, and differences in the scale of the images themselves, the classification of natural scene images is still difficult to deal with...

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
CPCG06F18/2411G06F18/214
Inventor 焦李成张丹马文萍屈嵘曾杰刘红英王爽侯彪杨淑媛尚荣华
Owner XIDIAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com