Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Small target semantic segmentation sun dark bar detection method combined with target detection

A technology for semantic segmentation and object detection, applied in the field of image processing, can solve the problems of missed and misjudged image detection accuracy, and achieve the effects of improved accuracy, improved detection accuracy, and improved edge detection accuracy.

Pending Publication Date: 2020-12-18
KUNMING UNIV OF SCI & TECH
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The object of the present invention is to provide a small target semantically segmented sun dark stripe detection method combined with target detection, which is used to solve the problem that the existing dark stripe detection method will make the intensity too low for images with uneven intensity on the Hα full sun surface image. The position of the position is recognized as a dark bar, and the intensity features of these positions have a strong similarity with the dark bar, which leads to the problems of missed and misjudged images and low detection accuracy in the network.

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
  • Small target semantic segmentation sun dark bar detection method combined with target detection

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0025] Step 1, prepare the data set, the full-helix image of the sun Hα comes from the Big Bear Lake Observatory in the United States. First, the image is preprocessed, and the median filter is used to reduce the noise of the image, and then the data set is enhanced with data enhancement technology. Finally, manual labeling is carried out, and it is divided into a training set and a verification set of the detection model in proportion;

[0026] In step 1, data enhancement is combined with color transformation, translation, scaling, rotation, etc., wherein the size of image scaling is set to 512*512, and color transformation randomly changes image brightness, saturation, and contrast. Use the labelme tool to label the dark bars of the sun in the full-helioscopic image. The Hα full sun image data set is mixed in equal proportions, and divided into a training set and a verification set for the detection model according to a ratio of 8:2. The training set is used for model train...

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 small target semantic segmentation sun dark bar detection method combined with target detection. The method comprises the following steps: firstly, preprocessing an Halpha full-solar-surface image, manually labeling the image, and dividing the image into a training set and a verification set of a detection model in proportion; constructing a semantic segmentation network,obtaining an overall semantic segmentation model through a data set training network, and selecting a training model with the best test set detection effect; cutting the small target image blocks according to a pixel-level real mark graph provided by the data set, and making a small target data set; training a target detection network through the small target detection data set obtained in the third step to obtain a target detection model; designing a semantic segmentation network for small target segmentation, and training the network by using the obtained small target segmentation data set;and finally, inputting the test image into the overall semantic segmentation model and the small target detection network. The method greatly improves the overall performance of a conventional sun dark strip recognition method, and improves the detection precision of weak and small dark strips and the edges of the dark strips.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a small target semantic segmentation sun dark stripe detection method combined with target detection. Background technique [0002] The sun's dark streak is a structure filled with low-temperature and high-density plasma that exists in the corona. There are no small differences between the dark bars, such as their appearance, size, and stable existence time. Some streaks take up half of the sun's surface, while others are just a tongue of flame at the edge of the sun's surface. In the observation of the full-helix solar image, because the dark stripe has a significant absorption effect on the background radiation of the bottom atmosphere, the dark stripe is lower than the corona temperature, so it appears as a slender dark black stripe structure under the solar background. Typically, dark streaks are cooler but denser than the surrounding corona. In the H-band observa...

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): G06T7/00G06T7/12G06T7/136G06N3/04G06N3/08G06T5/00
CPCG06T7/0002G06T7/12G06T7/136G06N3/08G06T2207/20081G06T2207/20084G06T2207/20132G06N3/045G06T5/90G06T5/70Y04S10/50
Inventor 尚振宏辛泽寰杨志鹏
Owner KUNMING UNIV OF SCI & 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
Eureka Blog
Learn More
PatSnap group products