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

Construction method of threshold learnable local binary network based on texture description and deep learning and classification method of remote sensing images

A local binary and deep learning technology, applied in the field of image processing, can solve problems such as poor model classification performance, and achieve the effect of solving poor model classification performance and improving performance

Active Publication Date: 2020-02-11
WUHAN UNIV
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] In view of this, the present invention provides a method for constructing a threshold learnable local binary network based on texture description and deep learning and a remote sensing image classification method to solve or at least partly solve the problem of model classification in existing methods in the prior art. Technical issues with poor performance

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
  • Construction method of threshold learnable local binary network based on texture description and deep learning and classification method of remote sensing images
  • Construction method of threshold learnable local binary network based on texture description and deep learning and classification method of remote sensing images
  • Construction method of threshold learnable local binary network based on texture description and deep learning and classification method of remote sensing images

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] This embodiment provides a method for constructing a threshold learnable local binary network based on texture description and deep learning, please refer to figure 1 , the method includes:

[0044] Step S1: Obtain a remote sensing image dataset, divide the dataset into two non-overlapping sub-datasets on average, and use them for training and cross-validation respectively.

[0045] Specifically, split the dataset D equally into two non-overlapping sub-datasets D t ,D v , used for training and cross-validation, respectively; the size of all dataset images is n×n pixels.

[0046] Step S2: Load the ResNet-50 network model pre-trained on the ImageNet data set, and modify the output dimension of the last fully connected layer of the ResNet-50 network model to the dimension corresponding to the image category, and use it in the remote sensing The stochastic gradient descent and backpropagation algorithms are used for fine-tuning on the image data set, and the ResNet-50 de...

Embodiment 2

[0077] Based on the same inventive concept, this embodiment provides a remote sensing image classification method, including: inputting the remote sensing image to be classified into the threshold value learnable local binary network TLBPNet based on texture description and deep learning constructed in Embodiment 1, and obtaining Image classification results.

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 construction method of a threshold learnable local binary network based on texture description and deep learning. The method comprises the following steps of firstly, takinga remote sensing image data set; then, loading a ResNet-50 network model which is pre-trained on an ImageNet data set; modifying an output dimension of a last full connection layer of the ResNet-50 network model into a dimension corresponding to the image category, and training the ResNet-50 network model on the remote sensing image data set; optimizing a manual feature LBP based on the thought ofdeep learning; obtaining an LBP method with a learnable threshold value, and then taking the LBP method with the learnable threshold value as an LBP layer to be connected in series with a ResNet-50 network model pre-trained on an ImageNet data set, so as to obtain a local network LBPNet with the learnable threshold value; and then connecting the ResNet-50 depth model converged on the data set with the LBPNet in parallel, and constructing a threshold learnable local binary network TLBPNet based on texture description and deep learning. The method can improve the classification performance of remote sensing images.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for constructing a threshold learnable local binary network based on texture description and deep learning, and a remote sensing image classification method. Background technique [0002] As a basic problem in the field of computer vision, image classification has always been an important research direction. The present invention is mainly aimed at classification of remote sensing scenes. Classification of remote sensing scenes is a challenging and open research problem. With the rise of remote sensing instruments, a large amount of satellite data has appeared in the field of remote sensing. Therefore, deep learning has gradually been introduced into the image classification of remote sensing scenes and has been widely used in It has received more and more attention in many applications, such as land cover classification, object detection, etc. The problem is ...

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): G06K9/62G06N3/08
CPCG06N3/08G06F18/24G06F18/214
Inventor 何楚张清怡刘新龙石紫珊
Owner WUHAN UNIV