Weak and small target detection method based on video satellite data identification characteristics

一种鉴别特征、卫星数据的技术,应用在遥感信息处理领域,达到提高检测精度、增大差异的效果

Active Publication Date: 2020-11-13
BEIJING UNIV OF CIVIL ENG & ARCHITECTURE
View PDF15 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The object of the present invention is to provide a kind of weak target detection method based on video satellite data discrimination feature, thus solve the foregoing problems existing in the prior art

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
  • Weak and small target detection method based on video satellite data identification characteristics
  • Weak and small target detection method based on video satellite data identification characteristics
  • Weak and small target detection method based on video satellite data identification characteristics

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0057] This embodiment provides a method for detecting weak and small targets based on video satellite data identification features, such as figure 1 shown, including the following steps:

[0058] S1, cutting the obtained video satellite images into image blocks of the same size, and inputting them into the VGG16 backbone network to obtain video satellite data image features;

[0059] S2. Carry out data labeling on the cropped image block, use the labeling area after the data labeling as the identification feature extraction range, and use the symmetrical semantic segmentation model and the self-encoding network model to extract the identification features of the target;

[0060] S3, using a top-down adjustment mechanism to fuse the image features of the video satellite data and the extracted target identification features;

[0061] S4, using the background in the segmented image segmented by the symmetrical semantic segmentation model in step S2, using the attention mechanis...

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 weak and small target detection method based on video satellite data identification features. According to the method, the identification features are fully utilized to improve the detection precision of a weak and small target. A symmetric semantic segmentation model and a self-coding network are adopted to extract identification features of a target, a top-down adjustment mechanism is utilized to perform data fusion on image features and the identification features of the target, then background enhancement is performed through an attention mechanism, and the contrast difference between the target and the background is further increased; a multi-scale semantic analysis strategy is introduced, and a pyramid model is adopted to extract weak and small targets in video satellite data. The method focuses on the extraction and introduction of weak and small target identification features, further increases the difference between the target and the background through employing an attention mechanism, and can greatly improve the detection precision of the weak and small target in a video satellite.

Description

technical field [0001] The invention relates to the technical field of remote sensing information processing, in particular to a weak and small target detection method based on identification features of video satellite data. Background technique [0002] In the current existing technology, there is already a mature technology for detecting weak and small targets using ground video data. The classic moving target detection methods mainly include frame difference method, background subtraction method and optical flow method. The feature descriptors used in the above method include: Haar features, SIFT features, and HOG features, etc.; classifiers include logistic regression, decision tree, Adaboost, and SVM, etc. The feature extraction algorithm of this type of method is highly dependent on the rationality of manual settings, and its generalization ability is weak, resulting in limited application scenarios. [0003] Using ground video data to detect weak and small targets,...

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): G06K9/00G06K9/34G06K9/46G06K9/62G06N3/04
CPCG06V20/13G06V10/267G06V10/462G06N3/045G06F18/25G06V10/82G06V20/64G06V20/41G06F18/24G06F18/2132
Inventor 吕京国曹逸飞曲宁宁扈廷锐
Owner BEIJING UNIV OF CIVIL ENG & ARCHITECTURE
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