Thermal sensors

https://github.com/OpenHD/Open.HD/wiki/Thermal-Cameras

Seine Meinung zu unserer Anwendung

Sensorspecs
	<10° Öffnungswinkel
	320x240 Auflösung
Quadratsensor vs Zeilensensor
	Quadratsensor besser weil sehr gutes preisleistungsverältniss / hohe stückzahlen
zeilensensor
	nur bei satelliten
	kosten höher
	nicht empfehlenswert laut ihm

Flir Sensoren

lepton 
	3.0 
		datenblat interface sensor
			SDI Interface selbst entwickeln	
		Kann keine Temperatur messen, nur Unterschiede anzeigen
	3.5 
		kann kan radiometrische information == absolute temperatur ausgeben
		für uns nicht relevant
		Wird in Coronatimes gerade sehr gebraucht für Messgeräte...
boson
	320 640
	besser für uns geeignet seiner meinung nach
	aber 10X so teuer

Temperaturunterschied ist entscheident

Von oben kuckend ist temepratur bei menschen nicht so hoch
Schräg wäre besser weil da mehr warme fläche vom menschen	

Lepton breakout board

V1.0 alte version , kann vlltpaar features nicht, aber hat keine beschwerden gehört, 30$
V2.0 60$

Erkennung mittels sensoren , 3 unterschiedliche stufen

Johnson criteria
	https://www.ecamsecure.com/blog/thermal-cameras/johnsons-criteria-for-thermal-camera-performance/
detektion 
	2 linienpaare aus = 4 pixel			
erkennung 
	6 linienpaare
identifikation / klassifikation 
	12 linienpaare

Defekte pixel

1-2% pixel tot
Müssen berücksichtigt werden
=> kann mittels tracking über zeit / bestätigung über zeit gelöst werden

1. 160×120 spacial res
2. 50 mKelvin temperature res
3. 71/56° opening angle
4. not waterproof
5. more info in this paper: https://www.vutbr.cz/www_base/zav_prace_soubor_verejne.php?file_id=181847
6. ~250€

Flight@300m

Flight@500m

• https://www.digikey.de/product-detail/en/flir-lepton/500-0726-01/500-0726-01-ND/6163867
  • Bought at acalbfi.com
• https://www.digikey.de/product-detail/en/flir-lepton/250-0577-00/1577-250-0577-00-ND/10385179
  • Bought v1.4 at acalbfi.com

Compared to the Lepton 3.5 it doesnt offer temperatur measurements, but we are not interested in it anyway so we can use the 3.0 version. Lepton 3.5 cannot be shipped in corona times anyway.

• Python Interfacing : https://github.com/groupgets/pylepton/tree/lepton3
  • Use lepton3-dev branch for lepton3.x support
  • Master got only lepton2.x support
• Live video: https://github.com/groupgets/LeptonModule/tree/master/software/raspberrypi_video
• Paper on the sensor: https://www.vutbr.cz/www_base/zav_prace_soubor_verejne.php?file_id=181847

* https://www.thingiverse.com/thing:1563825

• Better imho

* https://www.thingiverse.com/thing:2581533

• HT-201 / HT-301
• 320×240 pixel
• 32° Opening angle
• https://www.amazon.de/Thermal-Infrarot-W%C3%A4rmebildkamera-Mobiltelefone-Aufl%C3%B6sung/dp/B07Y4PJF2S/ref=sr_1_1?__mk_de_DE=%C3%85M%C3%85%C5%BD%C3%95%C3%91&dchild=1&keywords=ht-201+thermal&qid=1607201370&sr=8-1

• 206 x 156 Thermal Sensor
• 36° Field of View
• < 9 Hz Frame Rate
• 1,000 ft. Detection Distance (~300m)
• not waterproof!
• ~250€
• lensmaterial Chalcogenide

Flight@300m

Flight@500m

Flight@700m

• https://github.com/maartenvds/libseek-thermal

Needs special material as glas blocks IR. Common materials for LWIR windows include silicon, germanium, and zinc selenide (LWIR absorption in silicon is on the order of 15%/mm, which means NEDT is adversely affected using a silicon window. Bulk absorption in germanium and zinc selenide is negligible, and performance is essentially unchanged provided both surfaces of the window are anti-reflection (AR) coated.)

It should be stated that occasional water, be it fresh or salt, does not damage Germanium lenses if they are rinsed and dried afterwards. Fire fighting cameras regularly get drenched in water and the camera lens protector just needs to be gently cleaned and dried after the shout. from https://www.eevblog.com/forum/thermal-imaging/death-of-a-camera-by-drowning-!-flir-vue-its-vulnerabilities-and-death/msg1306467/#msg1306467

From FLIR Lepton Datasheet

More Info:

• Materials in general: https://www.eevblog.com/forum/thermal-imaging/buying-used-thermal-imaging-lenses-general-buying-guidance-from-fraser/
• Cheap Zinc Selenide ZnSe Lenses @ ebay: https://www.eevblog.com/forum/thermal-imaging/znse-lenses-on-ebay/
• Thermal transparnt plastics: https://miltechlabs.com/index.php?route=product/product&product_id=986
• Cheap PIR thermal transparent windows: https://www.kube.ch/produkte/pir-optics/pir-fenster/

Lepton 3.0 Germanium Lens

Bought here: https://de.aliexpress.com/item/32958470241.html

* Image

* Transmission of lens

* Without lens

Intensity of hottest point: 210

* With lens

Intensity of hottest point: 124

The lens dampens the intensity of the heat by at least ~40%, whis is roughly in line with the chart above.

It also created a lens center spot effect in the center of the image. We need to see if this might cause problems.

To be tested in real life conditions.

Test different materials for thermal transparancy with Lepton 3.0 @ Pi4

This time i take photos with normalized colors to min and max temperature. Actually a better test would be using absolute value images. Next time…

left to right

1. “thick foil” ~ like a paper 2. “superthin foil” 3. transparent ruler ~2mm 4. magnifying glass ~2mm

0. nothing

1.

2.

3.

4.

without

left to right

1. “superthin foil” 2. “thick foil” ~ like a paper 3. transparent ruler 2mm thick 4. glas

Its visible, that thin plastics are a option. The actual thinnest one doesnt protect at all, its too thin. But maybe we can find some cheap fitting material, or we use the professional thermal plastics.

• Images of water are more cold at center than edges
  • Due to vignetting
  • https://en.wikipedia.org/wiki/Vignetting

• Flat field calibration
  • Build avg frame over multiple frames and save flat_field_calib
  • frame_corrected = frame_raw - flat_field_calib
  • Implementation https://github.com/OpenThermal/libseek-thermal/blob/master/src/SeekCam.cpp#L101