Classification detection method for marine targets in SAR image based on convolution neural network

A convolutional neural network and target classification technology, applied in biological neural network models, neural architectures, instruments, etc., can solve problems such as unreasonable network structure, low algorithm practicability, and small number of samples, and achieve strong generalization ability, The effect of strong portability and simple structure

Inactive Publication Date: 2018-11-16
BEIHANG UNIV
View PDF3 Cites 19 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Many marine target classification algorithms based on convolutional neural networks have a small number of sample categories and s

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
  • Classification detection method for marine targets in SAR image based on convolution neural network
  • Classification detection method for marine targets in SAR image based on convolution neural network
  • Classification detection method for marine targets in SAR image based on convolution neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0080] This embodiment proposes a method for classifying and detecting SAR marine targets based on a convolutional neural network. This embodiment specifically includes the following steps:

[0081] Step 1: Perform power image conversion and quantization on the SAR single-view complex image to obtain;

[0082] Specifically:

[0083] According to the formulas (1) and (2), 60 GF-3 single-view complex images with a size of 16384×16384 can be used to obtain quantized SAR images, one of which is as Figure 4 As shown; a large number of training set samples and a suitable network structure are prerequisites for convolutional neural networks to achieve better classification results. The present invention provides a large number of training samples for the convolutional neural network by using, for example, a large number of SAR images of the Gaofen-3 radar satellite.

[0084] Step 2: Segment the high-resolution SAR image containing marine targets (that is, the quantized SAR image o...

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 classification detection method for marine targets in an SAR image based on a convolution neural network, which comprises the steps of first, converting an SAR single-look complex image containing marine targets into a power image, and carrying out quantization on the power image; second, segmenting the quantized power image, adding sample category labels, then constructing a training set, and establishing a data set; third, performing mean removing processing on the training set; fourth, building a convolution neural network model; fifth, training the convolution neural network model by using the training set; and sixth, inputting a slice to be classified into the convolution neural network model to obtain a classification detection result. The classification detection method has the characteristics of high reliability, strong generalization ability, low computational complexity, high practicability, wide application range and the like.

Description

technical field [0001] The invention relates to a method for classifying and detecting marine targets in SAR images based on a convolutional neural network, belonging to the fields of image processing and computer vision. Background technique [0002] Synthetic aperture radar originated in military applications and was developed in the 1950s as an active earth observation system. Synthetic aperture radar can conduct all-weather and all-weather observations of the earth without being affected by climate, weather, light and other conditions. It is widely used in: agriculture, soil moisture, forestry, geology, hydrology, flood and ocean monitoring, oceanography, Ship and oil slick detection, ice and snow detection, land cover mapping, altitude mapping, and earth change detection. [0003] In the past ten years, the convolutional neural network has been greatly developed. The convolutional neural network integrates feature extraction, feature selection, and feature classificati...

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): G06K9/00G06K9/34G06K9/62G06K9/46G06N3/04
CPCG06V20/13G06V10/267G06V10/40G06N3/045G06F18/214
Inventor 陈杰马梦原杨威李春升
Owner BEIHANG 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