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

RGB-D image classification method and system

A technology of RGB-D and RGB images, applied in the field of RGB-D image classification methods and systems, can solve the problems of low classification accuracy, insufficient classification of noisy data and high-similar images, and achieve strong representation ability and discrimination ability, classification Fast, overcoming the effect of noisy data

Active Publication Date: 2016-01-06
SOUTH CHINA AGRI UNIV
View PDF3 Cites 45 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] In order to overcome the insufficient classification of noise data and highly similar images and the problem of low classification accuracy, the present invention provides an RGB-D image classification method and system that integrates deep learning and group sparse representation, and automatically obtains low-level and middle-level features of the image. Overcome the classification defects of noisy data and highly similar images, use block dictionary optimization and group sparse coding methods to perform feature sparse representation, obtain RGB-D image high-level feature sparse representation and classify, and improve the accuracy of RGB-D image classification and recognition

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
  • RGB-D image classification method and system
  • RGB-D image classification method and system
  • RGB-D image classification method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0055] Such as figure 1 Shown is a flow chart of a specific embodiment of an RGB-D image classification method of the present invention. see figure 1 , the concrete steps of a kind of RGB-D image classification method of this specific embodiment include;

[0056] S101. Process the source RGB image and the Depth image respectively to extract low-level features.

[0057] In this step, the low-level feature extraction of the image is to use a single-layer convolutional neural network (CNN) to perform convolution operations on the RGB image and the Depth image respectively to complete the extraction of low-level features of the image, such as figure 2 with 3 As shown, the specific process of extraction is as follows:

[0058] S1011. Use K filters for each size d i (such as d i =148) image is convolved, and the size of the convolution block is d p , normalize and whiten the convolutional blocks, and finally form K feature filter maps, each feature map has m convolutional bl...

Embodiment 2

[0091] On the basis of Embodiment 1, the present invention also provides an RGB-D image classification system. A kind of RGB-D image classification system of the present invention specifically comprises:

[0092] The low-level feature extraction module uses CNN to process the source RGB image and the Depth image to extract low-level features;

[0093] The middle-level feature extraction module, through the recurrent neural network (RNN), performs feedback learning on the low-level features of the image, and learns the middle-level features of the image;

[0094] The high-level feature extraction module is used to adopt the intra-block constraint dictionary learning method to perform feature group sparse representation of the middle-level features of the image, and obtain the high-level feature representation of the RGB-D image;

[0095] The classification module is used to input the high-level features of the RGB-D image into the linear support vector machine SVM to complete ...

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 relates to an RGB-D image classification method and system. The method comprises the steps of: S1, utilizing a convolution neural network (CNN) to process a source RGB image and a Depth image respectively, and extracting low level characteristics; S2, utilizing a recursion neural network (RNN) to carry out feedback learning on the image low level characteristics, and extracting image middle level characteristics; S3, adopting a block interior constraint dictionary learning method, carrying out characteristic set sparse expression on the image middle level characteristics, and obtaining high level characteristics of the RGB-D images; and S4, inputting the high level characteristics of the RGB-D images into a linear SVM to complete the classified identification of the RGB-D images. According to the invention, automatic characteristic extraction of the images is realized, learning RGB-D image characteristic expressions can effectively distinguish classification of noise data from high similarity images, and the classification precision of the RGB-D images is improved; in addition, the linear SVM is utilized, and the image classification speed is improved.

Description

technical field [0001] The present invention relates to the field of pattern recognition and image classification, and more specifically, to a RGB-D image classification method and system. Background technique [0002] RGB-D image classification is a new field of pattern recognition technology emerging in recent years. Compared with color image classification, 3D image classification with depth information can directly reflect the 3D characteristics of the object surface, and can overcome the shortcomings of color image classification that are susceptible to interference from factors such as illumination changes, shadows, object occlusion, and environmental changes, and integrate depth information. Image classification is becoming a current research hotspot. The emergence of Kinect camera provides the possibility to obtain RGB-D (RGB image + depth image) images with depth information economically and quickly. Domestic researchers have used images with depth information to ...

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): G06K9/46G06K9/62
CPCG06V10/513G06V10/56G06F18/2411
Inventor 涂淑琴薛月菊胡月明梁云
Owner SOUTH CHINA AGRI 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