CN111695529B - X-ray source detection method based on human skeleton key point detection algorithm - Google Patents
X-ray source detection method based on human skeleton key point detection algorithm Download PDFInfo
- Publication number
- CN111695529B CN111695529B CN202010544700.2A CN202010544700A CN111695529B CN 111695529 B CN111695529 B CN 111695529B CN 202010544700 A CN202010544700 A CN 202010544700A CN 111695529 B CN111695529 B CN 111695529B
- Authority
- CN
- China
- Prior art keywords
- network
- ray source
- telescope
- wide
- source detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
Abstract
The invention discloses an X-ray source detection method based on a human skeleton key point detection algorithm, which belongs to the technical field of celestial source discovery and identification and comprises the following steps: s10, constructing an X-ray source detection network based on a convolutional neural network; the X-ray source detection network comprises a positioning network, a local maximum value extraction algorithm, a key point region feature extraction algorithm and an intensity prediction network which are sequentially arranged; s20, training an X-ray source detection network by using a wide-field sky-patrol telescope observation image training sample; s30, inputting the observation image of the wide-field sky-patrol telescope to be identified into a trained X-ray source detection network to obtain a flow intensity predicted value of the observation image of the wide-field sky-patrol telescope to be identified. The invention can remove noise from the observed image to extract the data information of the source, support the extended source and has the source resolution capability similar to the distance; meanwhile, the positioning accuracy and recall rate on the test data can reach more than 99%.
Description
Technical Field
The invention belongs to the technical field of celestial body source discovery and identification, and particularly relates to an X-ray source detection method based on a human skeleton key point detection algorithm.
Background
The existing X-ray source detection algorithm is mainly a sliding cell method, after the sliding cell is established, a pixel corresponding to the center position of the cell is used as a signal value to scan an observed image, meanwhile, a pixel at the adjacent position (8 neighborhood) of the center point is marked as noise for calculating the signal to noise ratio, and if the signal to noise ratio calculated by the current sliding cell is larger than a given threshold value, the current position is marked as an X-ray source candidate position. The X-ray source candidate locations and a portion of the random background in the image are then subtracted, denoted "cheeses" map, the subtracted locations are filled in by an interpolation algorithm (typically bilinear interpolation) to generate a smoothed background map, and finally the scanning of the first step is repeated, but the interpolated background map is used to represent the noise intensity when calculating the signal to noise ratio, giving a list of X-ray sources.
The sliding cell method is fast to execute and does not depend on a priori assumptions, but has the core problem that the method is friendly to point sources only and has high confidence on bright sources only because the center pixel is taken as a candidate source, so that the series of methods do not have good identification capability for a plurality of complex extension sources.
Therefore, there is an urgent need for an X-ray source detection method that can extract source data information from an observation image by removing noise, can support extended sources, and has source resolution capability close to distance.
Disclosure of Invention
The invention aims to provide an X-ray source detection method which can exclude noise from an observation image to extract data information of a source, can support an extended source and has source resolution capability close to a distance, and adopts the following technical scheme:
an X-ray source detection method based on a human skeleton key point detection algorithm comprises the following steps:
s10, constructing an X-ray source detection network based on a convolutional neural network; the X-ray source detection network comprises a positioning network, a local maximum value extraction algorithm, a key point region feature extraction algorithm and an intensity prediction network which are sequentially arranged;
s20, training the X-ray source detection network by using a wide-view-field sky-patrol telescope observation image training sample; the training process comprises the following steps:
s21, inputting the training sample of the observation image of the wide-field sky-patrol telescope into the positioning network, and outputting a position probability heat map by the positioning network;
s22, inputting the position probability heat map into a local maximum extraction algorithm to obtain X-ray source coordinates;
s23, inputting the X-ray source coordinates into a key point region feature extraction algorithm to obtain an X-ray source region local map;
s24, inputting the regional local map into the intensity prediction network to obtain a flow intensity prediction value;
s25, when the flow intensity prediction precision meets the requirement, the X-ray source detection network training is completed;
s30, inputting the observation image of the wide-field astronomical telescope to be identified into the X-ray source detection network after training is completed, and obtaining a flow intensity predicted value of the observation image of the wide-field astronomical telescope to be identified.
Further, the key point region feature extraction algorithm is a human skeleton key point detection algorithm.
Further, the positioning network is a ResNet residual network, and the ResNet residual network comprises 4 ResBlock modules which are sequentially arranged.
Further, the local map of the region is a 9×9 local image centered on the X-ray source.
Further, the intensity prediction network comprises a dense full connection layer, a relu activation layer, a dense full connection layer and a sigmoid activation layer which are sequentially arranged.
Furthermore, the positioning network adopts a multi-stage loss function supervision mode, and each stage refines the positioning result of the previous stage.
Further, the wide field of view telescope observation image training samples comprise 80% training samples and 20% test samples.
The invention has the beneficial effects that:
the invention can remove noise from the observed image to extract the data information of the source, support the extended source and has the source resolution capability similar to the distance; meanwhile, the positioning accuracy and recall rate on the test data can reach more than 99%, the mean square error of the strength prediction and the true value is about 0.006, the astronomical observation can be served with high accuracy, and the problems that the X-ray source observation data is too complicated and difficult to effectively extract useful information under the big data background are solved.
Drawings
FIG. 1 is a diagram of the overall network architecture of the present invention
FIG. 2 is a schematic diagram of a positioning network
FIG. 3 intensity prediction network schematic
FIG. 4 position probability heat map of positioning network output
Fig. 59×9 partial images
Detailed Description
Example 1
An X-ray source detection method based on a human skeleton key point detection algorithm comprises the following steps:
s10, constructing an X-ray source detection network based on a convolutional neural network; the X-ray source detection network comprises a positioning network, a local maximum value extraction algorithm, a key point region feature extraction algorithm and an intensity prediction network which are sequentially arranged;
in this embodiment, the positioning network is a res net residual network, where the res net residual network includes 4 res block modules (as shown in fig. 2) arranged in sequence; the positioning network adopts a multi-stage loss function supervision mode, and each stage refines the positioning result of the previous stage.
The intensity prediction network includes a dense full connectivity layer, a relu activation, a dense full connectivity layer, and a sigmoid activation in sequence (as in fig. 3).
In this embodiment, the key point region feature extraction algorithm is a human skeleton key point detection algorithm.
S20, training an X-ray source detection network by using a wide-field sky-patrol telescope observation image training sample; the training process comprises the following steps:
s21, inputting training samples of observation images of the wide-field sky-patrol telescope into a positioning network, and outputting a position probability heat map (as shown in FIG. 4) by the positioning network;
s22, inputting the position probability heat map into a local maximum extraction algorithm to obtain X-ray source coordinates;
s23, inputting the X-ray source coordinates into a key point region feature extraction algorithm to obtain a region local map;
in this embodiment, the local view of the region is a 9×9 local image centered on the X-ray source (as in fig. 5).
S24, inputting the regional local graph into an intensity prediction network to obtain a flow intensity prediction value;
in this embodiment, the predicted value in the strength prediction process is the ratio of the strength of the extension plane to the strength of the center, so as to stabilize the training process.
S25, when the flow intensity prediction precision meets the requirement, the X-ray source detection network training is completed;
in this embodiment, the training samples of the observation image of the wide-field telescope comprise 80% of training samples and 20% of test samples.
In this embodiment, 10000 wide-field-of-view telescope observation images are selected as training samples, including 8000 training samples and 2000 test samples. The positioning network and the intensity prediction network are respectively trained, wherein the data set of the intensity prediction network is a local graph corresponding to all sources in the sample. When the two network test accuracy is stable and then training is completed, the positioning accuracy requirement in the example is as follows: the recall rate and the accuracy rate reach 99 percent, and the flow intensity prediction precision requirement is as follows: the mean square error of the intensity predictions and the true values is 0.006.
S30, inputting the observation image of the wide-field sky-patrol telescope to be identified into a trained X-ray source detection network to obtain a flow intensity predicted value of the observation image of the wide-field sky-patrol telescope to be identified.
The embodiment is deployed on a telescope data processing computer, taking an einstein probe satellite as an example, and the satellite is provided with a wide-field-of-view telescope and a follow-up observation telescope with higher precision.
In the embodiment, the observation image of the wide-field sky-patrol telescope is taken as input, the position and flow information of the X-ray source in the observation result are identified, the result is further fed back to a subsequent X-ray source classification algorithm (not provided by the invention), the source of the target type of interest designed in advance by scientific researchers is identified, and the follow-up observation telescope is utilized for high-precision tracking observation.
The foregoing description is only a preferred embodiment of the present invention, and is not intended to limit the technical scope of the present invention, so any minor modifications, equivalent changes and modifications made to the above embodiments according to the technical principles of the present invention still fall within the scope of the technical solutions of the present invention.
Claims (7)
1. An X-ray source detection method based on a human skeleton key point detection algorithm is characterized by comprising the following steps:
s10, constructing an X-ray source detection network based on a convolutional neural network; the X-ray source detection network comprises a positioning network, a local maximum value extraction algorithm, a key point region feature extraction algorithm and an intensity prediction network which are sequentially arranged;
s20, training the X-ray source detection network by using a wide-view-field sky-patrol telescope observation image training sample; the training process comprises the following steps:
s21, inputting the training sample of the observation image of the wide-field sky-patrol telescope into the positioning network, and outputting a position probability heat map by the positioning network;
s22, inputting the position probability heat map into a local maximum extraction algorithm to obtain X-ray source coordinates;
s23, inputting the X-ray source coordinates into a key point region feature extraction algorithm to obtain a region local map;
s24, inputting the regional local map into the intensity prediction network to obtain a flow intensity prediction value;
the predicted value in the strength prediction process is the ratio of the strength of the extension surface to the central strength;
the data set of the intensity prediction network is a local graph corresponding to all sources in the sample;
s25, when the flow intensity prediction precision meets the requirement, the X-ray source detection network training is completed;
s30, inputting the observation image of the wide-field sky-patrol telescope to be identified into the X-ray source detection network after training is completed, and obtaining a flow intensity predicted value of the observation image of the wide-field sky-patrol telescope to be identified.
2. The method of claim 1, wherein the keypoint region feature extraction algorithm is a human skeletal keypoint detection algorithm.
3. The method of claim 1, wherein the positioning network is a res net residual network, and the res net residual network includes 4 res block modules arranged in sequence.
4. The method of claim 1, wherein the local view of the region is a 9X 9 local image centered on the X-ray source.
5. The method of claim 1, wherein the intensity prediction network comprises a dense fully-connected layer, a relu activation, a dense fully-connected layer, and a sigmoid activation in sequence.
6. The method of claim 1, wherein the positioning network employs multi-stage loss function supervision, each stage refining the positioning results of the previous stage.
7. The method of claim 1, wherein the wide field of view telescope observation image training samples comprise 80% training samples and 20% test samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010544700.2A CN111695529B (en) | 2020-06-15 | 2020-06-15 | X-ray source detection method based on human skeleton key point detection algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010544700.2A CN111695529B (en) | 2020-06-15 | 2020-06-15 | X-ray source detection method based on human skeleton key point detection algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111695529A CN111695529A (en) | 2020-09-22 |
| CN111695529B true CN111695529B (en) | 2023-04-25 |
Family
ID=72481253
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010544700.2A Active CN111695529B (en) | 2020-06-15 | 2020-06-15 | X-ray source detection method based on human skeleton key point detection algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111695529B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1766591A (en) * | 2004-09-21 | 2006-05-03 | 通用电气公司 | System and method for an adaptive morphology X-ray beam in an X-ray system |
| CN101166996A (en) * | 2005-02-28 | 2008-04-23 | 先进燃料研究股份有限公司 | Apparatus and method for detection of radiation |
| CN102881003A (en) * | 2012-08-30 | 2013-01-16 | 暨南大学 | Method for removing cosmic rays in charge-coupled device (CCD) astronomic image |
| CN103971127A (en) * | 2014-05-16 | 2014-08-06 | 华中科技大学 | Forward-looking radar imaging sea-surface target key point detection and recognition method |
| FR3047829A1 (en) * | 2016-02-12 | 2017-08-18 | Cie Nat Du Rhone | METHOD FOR ESTIMATING THE POSITION OF THE SOLAR DISK IN A SKY IMAGE |
| CN109064478A (en) * | 2018-07-17 | 2018-12-21 | 暨南大学 | A kind of astronomical image contour extraction method based on extreme learning machine |
| CN109683208A (en) * | 2019-01-25 | 2019-04-26 | 北京空间飞行器总体设计部 | A kind of adaptation space X radiographic source Accuracy Analysis method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10013764B2 (en) * | 2014-06-19 | 2018-07-03 | Qualcomm Incorporated | Local adaptive histogram equalization |
-
2020
- 2020-06-15 CN CN202010544700.2A patent/CN111695529B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1766591A (en) * | 2004-09-21 | 2006-05-03 | 通用电气公司 | System and method for an adaptive morphology X-ray beam in an X-ray system |
| CN101166996A (en) * | 2005-02-28 | 2008-04-23 | 先进燃料研究股份有限公司 | Apparatus and method for detection of radiation |
| CN102881003A (en) * | 2012-08-30 | 2013-01-16 | 暨南大学 | Method for removing cosmic rays in charge-coupled device (CCD) astronomic image |
| CN103971127A (en) * | 2014-05-16 | 2014-08-06 | 华中科技大学 | Forward-looking radar imaging sea-surface target key point detection and recognition method |
| FR3047829A1 (en) * | 2016-02-12 | 2017-08-18 | Cie Nat Du Rhone | METHOD FOR ESTIMATING THE POSITION OF THE SOLAR DISK IN A SKY IMAGE |
| CN109064478A (en) * | 2018-07-17 | 2018-12-21 | 暨南大学 | A kind of astronomical image contour extraction method based on extreme learning machine |
| CN109683208A (en) * | 2019-01-25 | 2019-04-26 | 北京空间飞行器总体设计部 | A kind of adaptation space X radiographic source Accuracy Analysis method |
Non-Patent Citations (3)
| Title |
|---|
| Zhuoxi Huo等.Point source detection performance of Hard X-ray Modulation Telescope imaging observation.《2015 national astronomical observation, Chinese academy of science and IOP publishing Ltd》.2015,全文. * |
| 王强等.基于X射线的盒装水饺异物自动检测与分类.《计算机辅助设计与图形学学报》.2018,(12),全文. * |
| 马志贤等.基于SVM的X射线天文图像点源探测算法.《上海师范大学学报(自然科学版)》.2016,(02),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111695529A (en) | 2020-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109840471B (en) | Feasible road segmentation method based on improved Unet network model | |
| CN111209887B (en) | SSD model optimization method for small target detection | |
| CN105528638A (en) | Method for grey correlation analysis method to determine number of hidden layer characteristic graphs of convolutional neural network | |
| CN112862774B (en) | Accurate segmentation method for remote sensing image building | |
| CN111861880A (en) | Image super-fusion method based on regional information enhancement and block self-attention | |
| CN113989261A (en) | Unmanned aerial vehicle visual angle infrared image photovoltaic panel boundary segmentation method based on Unet improvement | |
| CN116071660A (en) | Target detection method based on small sample | |
| Yang et al. | Visible-assisted infrared image super-resolution based on spatial attention residual network | |
| CN114820668A (en) | End-to-end building regular outline automatic extraction method based on concentric ring convolution | |
| CN111695529B (en) | X-ray source detection method based on human skeleton key point detection algorithm | |
| CN112101113B (en) | Lightweight unmanned aerial vehicle image small target detection method | |
| Ma et al. | CNN-TransNet: A Hybrid CNN-Transformer Network with Differential Feature Enhancement for Cloud Detection | |
| Shi et al. | A photometry pipeline for SDSS images based on convolutional neural networks | |
| Jiang et al. | Semantic segmentation network combined with edge detection for building extraction in remote sensing images | |
| CN115272242B (en) | YOLOv 5-based optical remote sensing image target detection method | |
| CN113657290B (en) | Snail collection and fine classification recognition system | |
| CN112529908B (en) | Digital pathological image segmentation method based on cascade convolution network and model thereof | |
| AlDahoul et al. | RGB-D based multimodal convolutional neural networks for spacecraft recognition | |
| CN111435537B (en) | Model training method and device and pose optimization method and device based on mosaic | |
| CN111435086B (en) | Navigation method and device based on splicing map | |
| CN113989612A (en) | Remote sensing image target detection method based on attention and generation countermeasure network | |
| Gao et al. | RoadSeg-CD: A network with connectivity array and direction map for road extraction from SAR images | |
| CN112668615A (en) | Satellite cloud picture prediction method based on depth cross-scale extrapolation fusion | |
| CN110443817A (en) | A method of improving image segmentation precision | |
| Guo et al. | A coarse to fine network for fast and accurate object detection in high‐resolution images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |