Ship target identification method based on feature migration

A target recognition and target technology, applied in the field of ship target recognition, can solve the problems of poor target recognition effect and achieve the effect of improving the effect

Active Publication Date: 2019-12-20
HARBIN INST OF TECH
View PDF18 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003]The purpose of the present invention is to solve the problem that in the existing methods, the target to be recognized is different from the target data of known training in terms of appearance and imaging quality. , leading to the problem of poor recognition of the target to be recognized, and a ship target recognition method based on feature transfer is proposed to be applied to typical remote sensing target 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
  • Ship target identification method based on feature migration
  • Ship target identification method based on feature migration
  • Ship target identification method based on feature migration

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0028] Specific implementation manner one: such as figure 1 Shown. The method for recognizing ship targets based on feature migration described in this embodiment includes the following steps:

[0029] Step 1. Select the high-resolution ship image as the training set image, and separately cut out the target slice of each image in the training set, and use the high-resolution ship image in the training set as the source domain to obtain the target slice of the source domain;

[0030] High-resolution ship image refers to the resolution within 0.5m;

[0031] Step 2: Calculate the HOG feature of the source domain target slice obtained in step 1, and vectorize the calculated HOG feature to obtain the vectorized HOG feature of the source domain (each slice has a corresponding vectorized HOG feature) ;

[0032] Step 3: For the low-resolution ship image to be recognized by the target, cut out the target slice of the image to be recognized, and use the low-resolution ship image to be recogniz...

specific Embodiment approach 2

[0045] Embodiment 2: The difference between this embodiment and the first embodiment is that in the first step, the size of the target slice is set according to the size of the target in the high-resolution ship image.

[0046] For the recognition of ship targets, the training set images are from the Google Earth data source. The targets in the training set images are divided into aircraft carriers, destroyers and cruisers, and the target slices corresponding to each training set image have the same size.

specific Embodiment approach 3

[0047] Specific embodiment three: this embodiment is different from specific embodiment two in that: the specific process of step four is:

[0048] Step 4: Generate subspace:

[0049] The vectorized HOG feature of the source domain and the vectorized HOG feature of the target domain are respectively subjected to standard normalization (mean value is 0, variance is 1), and the normalized vectorized HOG feature of the source domain and the normalization of the target domain are obtained. Vectorize the HOG feature after transformation;

[0050] After PCA (Principal Component Analysis) transformation is performed on the normalized vectorized HOG feature of the source domain, the feature vector corresponding to the first d large feature values ​​is selected, and the selected feature vector is used as the base B of the subspace of the source domain S ;

[0051] In the same way, after PCA transformation is performed on the normalized vectorized HOG feature of the target domain, the feature v...

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 ship target recognition method based on feature migration, and belongs to the field of ship target recognition. According to the method, the problem of poor recognition effect of the to-be-recognized target due to the fact that the to-be-recognized target is different from the known trained target data in the aspects of appearance and imaging quality in an existing methodis solved. According to the invention, HOG features of ship images with different resolutions are extracted; and, according to the transfer learning method based on spatial alignment and probabilityadaptation, the HOG features of the source domain and the HOG features of the target domain are mapped to the same feature space, then probability adaptation and instance weight adjustment are carriedout in the same feature space, and a new source domain vectorization HOG feature and a new target domain vectorization HOG feature are regenerated, the support vector machine is trained by using thenew source domain vectorization HOG feature, and target recognition is performed on the to-be-recognized image by using the trained support vector machine. The method can be applied to identificationof the ship target in the remote sensing image.

Description

Technical field [0001] The invention belongs to the field of ship target recognition, and specifically relates to a method of ship target recognition based on feature migration. Background technique [0002] For optical sensors, due to changes in observation position, height, etc., the resolution of the image obtained by the target will also change. As the resolution of the images obtained by the same target changes, they often follow different distributions. For traditional machine learning methods, it is usually assumed that the training data and the test data meet the same distribution, but in fact, the target that needs to be recognized in the optical remote sensing image is often the same as the known training target data in appearance, imaging quality and other characteristics The aspects are different, which leads to the inability to identify the target well, and the effect of identifying the target to be recognized is poor. Therefore, how to improve the target recogniti...

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): G06K9/00G06K9/46G06K9/62
CPCG06V20/00G06V10/50G06V2201/07G06F18/2411G06F18/214
Inventor 陈浩郭斌李宏博高通
Owner HARBIN INST OF TECH
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