Research on low-contrast wounded target search technology based on hyperspectrum
author: Gavin
2022-01-07
Quickly and effectively search for the wounded is to reduce the disappearance and death rate The primary factor. Judging from the characteristics of technological development and the way of use, the search for the wounded Technology can be divided into contact and non-contact. Existing contact life-saving equipment mainly There are effects such as inconvenient use of the wearable device and increased body load. And Natural disasters and other accidents are sudden, urgent, and unpredictable Etc. It is impossible for the wounded to carry life-saving equipment with them anytime and anywhere before the accident, And when the accident happened, the injured fell into a coma and could not trigger the equipment or the life-saving equipment was affected. Damage, poor communication signal quality, etc., will bring unknown causes to rescue work. White. The non-contact search technology mainly uses the vital characteristics of the human body to search. No need to wear any equipment to avoid the above problems.At present, non-contact search technologies at home and abroad can be divided into acoustic, optical, odor, microwave signals and other technologies. Among them, optical signals should be the most widely used. UAVs equipped with high-definition cameras and thermal infrared imagers are used for low-altitude searches. In a low-contrast environment, that is, when the casualty's clothing color is similar to the background color, it is difficult to find the casualty by relying on the airborne visual load. Hyperspectral target recognition technology has the advantages of anti-camouflage and anti-interference. It has its unique advantages to search for the wounded in a low-contrast environment.
Principle
Hyperspectral technology expands the imaging information from visible light to the 400-1000 nm band, uses prior knowledge to analyze the known spectrum curve, and recognizes and classifies by comparing the matching degree of the spectrum of each pixel in the hyperspectral image with the known spectrum. At present, hyperspectral imaging technology has been widely used in agriculture, forestry, ecological and environmental protection and other fields. Small drones are used to carry hyperspectral cameras for low-altitude patrols.
Aviation can realize small-scale high-precision remote sensing monitoring. The main principle is to affect the reflectance spectrum changes of the canopy population according to the content of chlorophyll and water in the vegetation, thereby reflecting the growth of the vegetation, diseases and insect pests, and water stress conditions in the area. Hyperspectral technology can effectively identify "metamer" targets. Research has found that hyperspectral can quickly and effectively identify camouflaged targets that cannot be identified by traditional visible light images. The spectrum, intensity, and polarization state of the target are integrated to improve target detection and recognition. Ability to analyze the camouflage effects of camouflaged targets in different backgrounds using hyperspectral images acquired in the visible and near-infrared bands.
Solution
The hyperspectral imaging system used in the experiment is composed of a hexarotor UAV, ATH1010 hyperspectral imager, ground control system, stabilized gimbal, standard reflectance plate, etc. ① ATH1010 hyperspectral camera [Optosky] adopts a slit design at the incident end , The band range is 400~1000nm, 300 channels, 1920×1080 pixel CCD detector, high spectral resolution is less than 2.6nm, provide data support for subsequent experimental analysis. ② Hexa-rotor UAV, can carry 6 kg payload and fly for 45 minutes; the maximum Sailing speed is 6m/s.
Using hyperspectral imaging technology to collect remote sensing images of ground camouflage casualties, and simulating blood spray camouflage clothing to simulate casualties in low-contrast environments. After hyperspectral remote sensing images are preprocessed, the target spectrum and background spectrum are extracted, using hybrid tuned matched filtering and minimum energy constraints ( Constrained Energy Minimization, CEM) and Adaptive Coherence Estimator (ACE) are three common hyperspectral feature recognition methods for casualty identification.
Test result
Choose sunny and breezy weather and test on the school football field. The measurement area is 90 m×120 m, with yellow and green grass as the main area. Camouflage uniforms are laid on the grass and artificial blood is sprayed to disguise as the wounded.
After the image is preprocessed, the target spectrum and background spectrum are extracted, and more than 100 pixels are selected for each category to statistically analyze the spectrum curve. As shown in the figure, the spectrum of various ground objects has large differences, and the method of spectrum recognition can be used for target recognition.
After the preliminary recognition results are obtained, convolution filtering and minimum clustering methods are performed to remove isolated small spots in the results to reduce misrecognition. The final recognition result is shown in the figure. The red pixels in the figure are the recognition targets, and the outlines of the 6 disguised wounded can be clearly seen
Conclusion
The basic principle of hyperspectral target recognition is to calculate the similarity between the spectrum curve of an unknown object and the spectrum curve of a known object, and to set the corresponding threshold according to the degree of similarity to identify and classify the object. When extracting the target spectrum, the selected pixels are all pure pixels. When the blood stains in the mixed pixel are relatively small, the difference between the target spectrum and the target spectrum is large. Therefore, the target mixed pixel cannot be identified when the threshold is set small. In addition, it is also related to the characteristics of bloodstains. Bloodstains are semi-transparent objects. Mixing with different ground objects and changes in the degree of dryness and humidity will affect the changes in spectral characteristics. However, when counting the number of target pixels, all pixels containing blood are included in the target pixels, resulting in a low recognition accuracy rate. Although the edges of the bloodstains were not recognized, the area of the bloodstains became smaller and the edges were jagged, but the outline of the bloodstains could still be clearly identified.
Related products
Using hyperspectral imaging technology to collect remote sensing images of ground camouflage casualties
Hyperspectral Camera:ATH1010
Airborne Hyperspectral Remote Sensing System:ATH9010
Drone Hyperspectral Imaging System:ATHL9010
Related articles
[1] de Oliveira D,Wehrmeister M.Using deep learning and low-cost rgb and thermal cameras to detect pedestrians in aerial images captured by multirotor UAV[J].Sensors,2018,18(7):2244.
[2] Lygouras E,Santavas N,Taitzoglou A,et al.Unsupervised human detection with an embedded vision system on a fully autonomous UAV for search and rescue operations[J].Sensors,2019,19(16):3542.
[3] Sun J,Li B,Jiang Y,et al.A Camera-based target detection and positioning UAV system for search and rescue (SAR) purposes[J].Sensors,2016,16(11):1778.
[4]Zamanallah M,Vergara O,Araus JL,et al.Unmanned aerial platform-based multi-spectral imaging for field phenotyping of maize[J].Plant Methods,2015,11(1):35．
Principle
Hyperspectral technology expands the imaging information from visible light to the 400-1000 nm band, uses prior knowledge to analyze the known spectrum curve, and recognizes and classifies by comparing the matching degree of the spectrum of each pixel in the hyperspectral image with the known spectrum. At present, hyperspectral imaging technology has been widely used in agriculture, forestry, ecological and environmental protection and other fields. Small drones are used to carry hyperspectral cameras for low-altitude patrols.
Aviation can realize small-scale high-precision remote sensing monitoring. The main principle is to affect the reflectance spectrum changes of the canopy population according to the content of chlorophyll and water in the vegetation, thereby reflecting the growth of the vegetation, diseases and insect pests, and water stress conditions in the area. Hyperspectral technology can effectively identify "metamer" targets. Research has found that hyperspectral can quickly and effectively identify camouflaged targets that cannot be identified by traditional visible light images. The spectrum, intensity, and polarization state of the target are integrated to improve target detection and recognition. Ability to analyze the camouflage effects of camouflaged targets in different backgrounds using hyperspectral images acquired in the visible and near-infrared bands.
Solution
The hyperspectral imaging system used in the experiment is composed of a hexarotor UAV, ATH1010 hyperspectral imager, ground control system, stabilized gimbal, standard reflectance plate, etc. ① ATH1010 hyperspectral camera [Optosky] adopts a slit design at the incident end , The band range is 400~1000nm, 300 channels, 1920×1080 pixel CCD detector, high spectral resolution is less than 2.6nm, provide data support for subsequent experimental analysis. ② Hexa-rotor UAV, can carry 6 kg payload and fly for 45 minutes; the maximum Sailing speed is 6m/s.
Test result
Choose sunny and breezy weather and test on the school football field. The measurement area is 90 m×120 m, with yellow and green grass as the main area. Camouflage uniforms are laid on the grass and artificial blood is sprayed to disguise as the wounded.
The basic principle of hyperspectral target recognition is to calculate the similarity between the spectrum curve of an unknown object and the spectrum curve of a known object, and to set the corresponding threshold according to the degree of similarity to identify and classify the object. When extracting the target spectrum, the selected pixels are all pure pixels. When the blood stains in the mixed pixel are relatively small, the difference between the target spectrum and the target spectrum is large. Therefore, the target mixed pixel cannot be identified when the threshold is set small. In addition, it is also related to the characteristics of bloodstains. Bloodstains are semi-transparent objects. Mixing with different ground objects and changes in the degree of dryness and humidity will affect the changes in spectral characteristics. However, when counting the number of target pixels, all pixels containing blood are included in the target pixels, resulting in a low recognition accuracy rate. Although the edges of the bloodstains were not recognized, the area of the bloodstains became smaller and the edges were jagged, but the outline of the bloodstains could still be clearly identified.
Related products
Using hyperspectral imaging technology to collect remote sensing images of ground camouflage casualties
Hyperspectral Camera:ATH1010
Airborne Hyperspectral Remote Sensing System:ATH9010
Drone Hyperspectral Imaging System:ATHL9010
Related articles
[1] de Oliveira D,Wehrmeister M.Using deep learning and low-cost rgb and thermal cameras to detect pedestrians in aerial images captured by multirotor UAV[J].Sensors,2018,18(7):2244.
[2] Lygouras E,Santavas N,Taitzoglou A,et al.Unsupervised human detection with an embedded vision system on a fully autonomous UAV for search and rescue operations[J].Sensors,2019,19(16):3542.
[3] Sun J,Li B,Jiang Y,et al.A Camera-based target detection and positioning UAV system for search and rescue (SAR) purposes[J].Sensors,2016,16(11):1778.
[4]Zamanallah M,Vergara O,Araus JL,et al.Unmanned aerial platform-based multi-spectral imaging for field phenotyping of maize[J].Plant Methods,2015,11(1):35．
Precision fertilization by UAV for rice at tillering stage in cold region based on hyperspectral remote sensing prescription map 
New product|recommendation-GF900 UAV-borne laser methane telemetry system 
Application of UAV-hyperspectral imaging for rice growth monitoring 
Application of hyperspectral camera in industrial inspection - Estimation Scheme of Rice Yield Based on UAV Hyperspectral Images
Surveillance of pine wood nematode disease based on satellite remote sensing images 
UAV forest fire patrol protection plan 
Application of Hyperspectral Imager in Disguised Target Recognition(part 2) 
Application of hyperspectral imager in detection of exogenous pests in jujube fruit 
Application of UAV Hyperspectral in Garbage Sorting 
Application of Hyperspectral Imager in Disguised Target Recognition(part 1) 
What is the hyperspectral imaging? 
Study on the degree of damage to the Asian carlocust with hyperspectral remote sensing model 
Nitrogen detection in cotton leaves based on hyperspectral 
FAQ_HYPERSPECTAL IMAGER FAQ 
Using ATHL9010 to estimate above-ground carbon stocks at the single-tree scale in subtropical forests - Airborne Hyperspectral Imaging in Early Monitoring of Pine Wood Nematode
A Study on Eutrophication of Lake Based on Hyper-spectraI Remote 
Application of Hyperspectral in the Construction of Hongshan Polymetallic Prospecting Model - Vegetation Classification of Alpine Grassland in Qinghai Lake Basin Based on HSI hyperspectral remote sensing data