EEVblog® Electronics Community Forum
Products => Thermal Imaging => Topic started by: Fraser on December 11, 2021, 04:04:01 pm
-
UPDATE : See post 32 in this thread before buying one of these cameras !
I have been watching some sales of reasonably priced thermal imaging cameras on eBay UK recently. Many will already know about the Farnell Multicomp branded UTi80P units that I posted about. There is another interesting camera available now and at around £60 to £70 seems to be a good deal.
The camera in question is the Hikvision DS-2TP31B-3AUF that was introduced for fever detection deployments.Now I would be the first to warn that the use of such a camera for fever detection is fraught with issues of accuracy but the camera is still a thermal imaging system so is worthy of a mention here.
An initial look at the specifications shouts “FLIR Lepton 3.5” to me but I have not investigated further yet. The bad news is that he advertising for this camera indicatedps a measurement range of only 30C to 45C with a claimed accuracy of +/-0.5C. That measurement accuracy claim has name looking at this product sideways as that is quite a claim when not using a known accurate temperature. Reference in the thermal scene. This may suggest that a very careful measurement calibration table has been employed in the camera over this limited temperature range in order to improve accuracy over generic calibration but I do not know for sure.
Interestingly, the pictures that I have see show a temperature scale on the right of the screen that extends far below the stated 30C minimum so we know that he core s autoranging as per any other camera. It is the temperature measurement system that may be limited in its measurement range. The use menu does show the temperature measurement range of 30C to 45C but there is no mention of an ability to change this.
So we have a thermal imaging camera that likely uses a Lepton 3.5 core and provides decent 160 x 120 pixel imaging that is upscale to 320 x 240 pixels on the LCD display. It MAY be locked to a limited temperature measurement range but it M@y also be possible,E to change that range. The thermal I,aging appears to be be of the conventional auto span variety to cover all temperatures within a scene.
So for around £70 you would be taking a risk on such a unit but it still looks a good deal to my eyes :) These units are normally very expensive and even if they are temperature measurement limited, they are still a Real thermal imaging camera and it may be possible to further analyse saved files. So many questions and sadly no answers yet. I will have to do some digging.
Fraser
-
Test-Meter UK sell these for over £1000 each +Vat :o But they do give you a free tripod or IR thermometer though :-DD
https://www.test-meter.co.uk/hikvision-ds-2tp31b-3auf-fever-screening-thermal-camera?utm_source=google_shopping&gclid=CjwKCAiAtdGNBhAmEiwAWxGcUi2MtyvvPNQw5ptsj6u41p3TcWHc5KHDPEtFRT3JmydWLmnHDIEYXBoCqCgQAvD_BwE (https://www.test-meter.co.uk/hikvision-ds-2tp31b-3auf-fever-screening-thermal-camera?utm_source=google_shopping&gclid=CjwKCAiAtdGNBhAmEiwAWxGcUi2MtyvvPNQw5ptsj6u41p3TcWHc5KHDPEtFRT3JmydWLmnHDIEYXBoCqCgQAvD_BwE)
Video is here...
https://youtu.be/r1h1erZQdyg (https://youtu.be/r1h1erZQdyg)
-
DS-2TP31B-3AUF User Manual attached from here:
https://us.hikvision.com/en/system/files_force/manual/ud13441b-b_baseline_handheld_thermography_camera_user_manual_of_ds-2tp31b-3auf_200509.pdf?download=1
-
The user manual for the standsrd (non fever) version...
https://www.hikvision.com/content/dam/hikvision/products/S000000001/S000000313/S000000317/S000002536/OFR003799/M000034794/User_Manual/UD18507B_Baseline_Thermographic-Handheld-Camera_User-Manual_200226.pdf (https://www.hikvision.com/content/dam/hikvision/products/S000000001/S000000313/S000000317/S000002536/OFR003799/M000034794/User_Manual/UD18507B_Baseline_Thermographic-Handheld-Camera_User-Manual_200226.pdf)
Fraser
-
Hmmm, interesting discovery....
https://www.aliexpress.com/item/1005001422640955.html (https://www.aliexpress.com/item/1005001422640955.html)
So the topic of this thread MAY be the same as the Hikvision H10 with a modified temperature measurement range ?
I see that the specification states 160 x 120 pixels at 25fps so these Hikvision cameras do not use a FLIR Lepton. This is actually good news as the recent microbolometer releases from China have impressed me.
Fraser
-
A close-up lens has been designed and offered for sale for the H10 model.....
https://www.isecuseshop.com/product/micro-lens-for-thermal-imaging-camera-hikvision-h10-h11/ (https://www.isecuseshop.com/product/micro-lens-for-thermal-imaging-camera-hikvision-h10-h11/)
https://youtu.be/x7jB-3sD4jc (https://youtu.be/x7jB-3sD4jc)
-
A video 'walk' through the Hikvision fever cam menu's.....
https://www.youtube.com/watch?v=r0Jxk-QAtQY (https://www.youtube.com/watch?v=r0Jxk-QAtQY)
Fraser
-
I have been reading about Hikvision and their capabilities. The core in this camera is likely from their own fabrication line.
An excerpt from an article about one of their Circa 2016 thermal security cameras……
————————
“ Hikvision’s competitors in this specialty include: Dahua and Uniview, both based on China, Bosch (Germany) and Axis (Sweden).
Setting Hikvision apart, however, is that “the company is able to design and manufacture its own products,” according to System Plus. The Chinese company has its own MEMS production line, MEMS packaging/tests, numerous surface-mount technology (SMT) lines and final assembly capabilities.
Intel, Hikvision and Movidius (now an Intel company) supply three key components unique to Hikvision’s camera:
Intel’s reconfigurable solution was responsible for digital signal processing after digitalization of Microbolometer output, and thermal management along with Maxim Peltier controller.
Hikvision itself designed a chip to do image signal processing, video encoding & encryption, ethernet interface.
Movidius processor enabled vision processing for AI applications.”
————————-
Fraser
-
A comparison of the standard model (H10) camera and the FLIR TG267. I am not a fan of The videos from TE and this video completely misses that the TG267 uses a standard single pixel IR thermometer for the measurement side of things.
At the end of the, video there is an image comparison that is interesting to see.
https://youtu.be/J3rh0C4LJRk
Fraser
-
A nice Hikvision sales video for the DS-2TP31 series :-+
https://youtu.be/tcU1FxJmkiY
Fraser
-
Video detailing the Fever detection camera working and menu options.
https://youtu.be/pwEYUW6mulE
Fraser
-
The standard version (non fever detection) specifications…
https://www.hikvision.com/content/dam/hikvision/products/S000000001/S000000313/S000000317/S000000320/OFR000431/M000008273/Data_Sheet/Datasheet-of-DS-2TP31-3AUF-20190531.pdf (https://www.hikvision.com/content/dam/hikvision/products/S000000001/S000000313/S000000317/S000000320/OFR000431/M000008273/Data_Sheet/Datasheet-of-DS-2TP31-3AUF-20190531.pdf)
Note that the unit uses a 17um pixel microbolometer :-+
Fraser
-
But it is not a "deal" yet, the camera is on an auction and more bids may come.
-
I have been in touch with the seller and he has more of these cameras and is open to offers on them. I understand that £60 will secure one.
It should be clearly understood that they are the fever detection version and we do not know whether the configuration permits a wider measurement range than 30C to 45C. Whilst Fever detection cameras may have commanded a high price due to their specialist nature, this also acts against them for generic use. Hence the low price that they seem to sell for on eBay.
Fraser
-
An initial look at the specifications shouts “FLIR Lepton 3.5” to me but I have not investigated further yet. The bad news is that he advertising for this camera indicatedps a measurement range of only 30C to 45C with a claimed accuracy of +/-0.5C. That measurement accuracy claim has name looking at this product sideways as that is quite a claim when not using a known accurate temperature. Reference in the thermal scene. This may suggest that a very careful measurement calibration table has been employed in the camera over this limited temperature range in order to improve accuracy over generic calibration but I do not know for sure.
I do not think they are smarter than Flir. And what Flir's reference procedure for screening is to take several (up to 10 if i remember correctly) reference samples, e.g. by sampling healthy peoples' skin temperature, likely averaging them, before beginning to screen.
Or this camera specs may simply be be bogus, like millions other products specs crafted in China.
-
No argument from me.I worked with a manufacturer who wanted +/-0.5C measurement accuracy for their new thermal camera design. I advised them that such would need a very carefully designed imaging core and a temperature reference in the thermal scene if the measurements were to be considered accurate. There are a lot of variables in a camera and thermal scene that make +/-0.5C guaranteed accuracy very hard to achieve. I have tested my FLIR thermal cameras against Black Bodies of known accuracy and had readings on the cameras within 0.1C of the Black Body setting. But that is a high quality, very expensive thermal camera and a Black Body test source….i.e. about as good as a measurement situation can get. Those same high quality FLIR cameras have a measurement specification of +/-2C or 2% (whichever is greater). FLIR know that the measurement accuracy can be influenced by many factors and reflect this in their specification. It is still recommended that a high quality industrial thermal camera has a calibration check every year as components can age or fail over time. Professional thermographers needing very accurate measurements will invest in a precision Black Body and check their cameras accuracy before a series of measurements.
In China, the Government advisers set a measurement accuracy requirement on thermal cameras of +/-0.5C for them to be acceptable for Fever Detection. Hence why this Fever Detection camera shows that exact specification. We do need to consider that a general use thermal imaging camera uses few calibration points across its wide measurement range so there is variance in the cameras accuracy over that range. If you limit a cameras measurement range to just that required and then generate many calibration points, the resulting measurement accuracy of that camera over the restricted temperature range will be better than that of the same camera that has a general use calibration. It is all about the measurement stability of the camera and accuracy of the calibration curve installed in the camera for radiance to temperature units conversion.
Fraser
-
It is all about the measurement stability of the camera and accuracy of the calibration curve installed in the camera for radiance to temperature units conversion.
Fraser
and measuring the flag to that accuracy, whether the flag actually is that temperature, whether the flag *looks* like that temperature, the temperature of the optics, the emissivity of the target surface, what might be reflecting off the target surface ........
Not to mention that for a virus that transmits before you have any symptoms it is fairly useless. You may well pick up a few very hot people, but will certainly let through a lot of infected cold people.
Bill
-
Or this camera specs may simply be be bogus, like millions other products specs crafted in China.
Sadly the usual 'we will tell you whatever you want to hear' approach which to be fair is the main way to survive in a totalitarian regime. Then lies defended to the last regardless of evidence to 'save face'.
"Are these fixings marine grade A4 stainless steel like the drawing says ?"
"Yes"
"Really ?"
"yes .. here is the certificate"
(idential to the material certificate from 4 years ago *we* sent *them*)
"OK so why are they now so rusty ?"
"............ well definitely Chinese equivalent standard so cannot be rusty"
"see picture"
silence................ price up by 4x
..........and repeat.........
-
Doing my research on this camera, I think it is best suited to those who want a cheap thermal imager that is not going to need measurement capability outside of the specified 30C to 45C. It could be quite an accurate camera within that range though.
This ‘B’ version of the DS-2TP31 camera series is running a bespoke firmware and I managed to download a firmware update file fir it from Hikvision. The file is a binary but I have not I have not identified the chipset that it is for. Most likely an ARM based SOC though.
It is unlikely to be a simple modification to revert the camera to standard DS-2TP31 specification and I could not find a standard version firmware upgrade file. If such were found, it might make reversion to standard specification possible but would require the cameras ID to be modified in order fir the firmware to be fooled into installing onto an ‘incorrect’ platform. There is also the issue of calibration. The ‘B’ model likely has a very different calibration table to the standard camera and that may, or may not result in calibration issues if a standard firmware is installed.
This camera is a very good example of what we may see in the future…… surplus Fever a detection cameras, both new and used, appearing on eBay after organisations dump them into disposal auctions or electronics recyclers. I have a suspicion that these Hikvision cameras are part of an aborted attempt to combat Corona Virus and have been sent for disposal by either a wholesaler or end user company. Farnell, CPC, Rapid and Radio Spares are all possibilities. The seller appears to have at least 12 drom the numbering he has used in past auctions on eBay. I bought the UTi80P in Multicomp branding last month from a chap who had a box of 10. Classic disposal stock sold by the box rather than individually.
Not all Fever Detection cameras will be the same in terms measurement temperature range but potential buyers need to consider that in order to meet some Government specifications, it may have been necessary to ‘tune’ the cameras performance and calibration to the human medical temperature range. Whilst this may produce an affordable thermal camera that performs very well at temperatures in the range ~30C to ~45C, that may mean that they cannot offer measurement outside that range. In order to modify such a thermal camera, the buyer would need to reverse engineer the operating system and hardware to identify the System on a Chip that is commonly used in these budget cameras. Obtaining firmware updates can be helpful in this task as some effectively replace the operating system and leave only the Calibration files unchanged. I am not a coder so cannot comment on the challenges and required techniques to ‘hack’ one of these Fever Detection cameras to a more useful general temperature measurement range. It would likely take some time to achieve and so may be better to buy a standard version of a Fever Detection camera in the first place :-//
I highlighted the auction for these Hikvision thermal cameras only because some experimenters may relish the opportunity to buy a cheap 160 x 120 pixel complete camera and modify it to their needs. It is unlikely to be a ‘turn-key’ solution to the general thermal camera user community due to the restricted measurement range. That said, I have not seen a specification for the cameras ability to display a thermal scene between a minimum and maximum temperature. So the camera may be fine for displaying hot spots on PCB’s etc but if outside the measurement range, the user might not be able to actually see the temperature of a component. A 20mm diameter close-up ZnSe lens would be needed for PCB work, so add around $20 for that to the cost of ownership. The sensible HFOV works to the cameras advantage with its 160 x 120 pixel resolution.
So this is likely to be a project camera unless someone wants it for its intended purpose. Bear this in mind before buying any thermal Camera that is specifically made for Fever Detection and has a narrow measurement range. They may be sold at great prices on the surplus market but can you use them in your application ?
Still good value for £60 in my opinion though :-+ If they could be modified into camera traps they would make great thermal camera traps for wildlife monitoring or could be used as wildlife spotting cameras in their original specification :) Vets could likely make use of these cameras as well as they would be well suited to imaging horses limbs etc for inflammation related heating. In fact, thinking about it, these would be great for farmers and vets :-+
Fraser
-
Bill_W,
In my personal experience, I have seen the designers of a certain brand of Chinese thermal camera genuinely trying to achieve the Chinese Governments specification for a Fever Detection camera. They were genuinely crest fallen when I tested the prototype and declared it unfit for purpose. My testing revealed two very significant findings….the USA made imaging core was totally unsuitable in “stand alone mode” for use as a Fever Detection camera due to inaccuracy issues and the Chief designer of the cameras electronics had mistakenly thought that the accuracy of a thermal camera relied only upon the accuracy of the Black Body source used to set a 2 point calibration covering >100C span. He was not experienced in thermal camera design. It was therefore no surprise that a budget core married to a generic embedded processor board and calibrated using generic calibration techniques and an incorrect calibration routine, failed to meet the +/-0.5C measurement accuracy requirement. The whole design, from the Ground up was flawed in terms of its accuracy capability. If the manufacturer had opted to use a decent thermal reference emitter in the thermal scene, many of the issues could have been resolved and the required measurement accuracy achieved. It was the error in the calibration routine and lack of an ‘in-scene’ thermal reference at human body temperature that condemned the camera to general duties rather than Fever Detection.
I might add that, after reading my four reports, the company recruited an experienced new Chief engineer who studied my test reports, agreed with my findings and corrected as much as was possible with the extant hardware package that he had to work with. I can honestly say that this particular Chinese thermal imaging company took product performance seriously but I also witnessed the pressure that their staff were under to get a project finished and out the door as quickly as possible to recover the development investment. The commercial business world is a tough place to work in if you want perfection and there is profit or investment recovery involved.
That Chinese thermal camera manufacturer may be unusual in their good attitude towards product performance but I hope not as the recent arrival of new, and very nice, microbolometer cores could hail a new era in affordable high frame rate thermal imaging cameras outside of USA borders.
To support what Bill_W has stated, it is very common for company marketing teams to ‘skew’ or massage specifications in an attempt to impress the potential purchaser and this has been the case across the World for decades. Just think about so called “Hi-Fidelity” audio systems with impressive output Wattage figures that turn out to be “Peak Music Power Output” (PMPO) rather than RMS and in small print there is the awful Total Harmonic Distortion figure associated with the poor design of the power amplifier ! Old tricks, just new equipments to play them on :)
Fraser
-
Why would I want to go through this acrobatics? I've chosen simply not to buy anything priced more then $10 ftom China. And even $10 items only the ones i cant find anywhere else. Saved myself a lot of time and headache.
-
Bill_W,
I might add that, after reading my four reports, the company recruited an experienced new Chief engineer who studied my test reports, agreed with my findings and corrected as much as was possible with the extant hardware package that he had to work with. I can honestly say that this particular Chinese thermal imaging company took product performance seriously
That though shows the problem, any Chinese supplier who tries to do the right thing will be both late and seem overpriced compared to the 'cowboy' competitors and will be crowded out especially on Alibaba etc.
The same applies to the marine stainless steel example. Once the price went up for using the right materials the 'corporate shopping department' goes looking for another 'cheap' supplier to restart the whole cycle based on the assumption that what you put on the drawing is what will come in as long as there is a 'certificate of conformity' with a stamp on it (known to the cynical as the happiness certificate).
-
OK, after some research on the DS-2TP31B-3AUF, I can confirm that Hikvision have produced a bespoke firmware designed for accuracy of measurement in the field of Human body temperature monitoring. The camera is not just a standard DS-2TP31 model with restricted measurement range and more calibration points. The firmware uses multiple frame captures and image processing to increase the accuracy of measurements. For this reason the manufacturer states that for best accuracy the camera should be static on a tripod and not hand held. Hikvision have also released a PC based image analysis software package that may be downloaded for free from their site.The capabilities of that analysis package have yet to be investigated.
To return this specialist version of the DS-2TP31 series cameras to general use configuration would likely require replacement of the firmware, rather than a change in configuration. There is then the question of the calibration table design as this may still be tuned for the limited temperature range used in the Fever Detection camera design.
So that is about as much as I can extract from everything I have read about this camera from public domain documents. I have yet to find the Patent document though. The fact that a dedicated specialist firmware was needed to achieve anything like acceptable Fever Detection measurement accuracy does not surprise me at all.
Fraser
-
When investigating the likely technology within a thermal camera design, it can sometimes be useful to look for FCC documentation for the camera, or a near relative to it. In the case of the DS-2TP31B-3AUF, it does not have an entry in the FCC approval database as it does not contain an RF transceiver. There is a thermal camera model that may share the same technology with this camera though. It is the bigger brother DS-2TP23 model that does contain an RF transceiver so there are internal pictures of its stored in the FCC database.
I attach the internal images document. Note the microbolometer section.... it is reminiscent of a ULIS Pico product, but VOx rather than A-Si. That is not one of the more recent miniature cores and, having looked at other Hikvision product internal images I have not seen the use of a miniature imaging cores in their products yet. I do wonder what imaging core design is inside the DS-2TP31B03AUF camera.
-
For anyone still following this thread..... an interesting development
I obtained a test image and used it with the Hikvision analysis software. What it revealed makes me think this Fever Detection camera may still have something to offer general users but with a big Caveat .... namely, I cannot verify temperature accuracy outside of the stated 30C to 45C calibrated range. That said, the Radiance to Temperature units conversion curve for such a thermal camera is normally pretty linear so a good calibration at 30c to 45C is a decent basis for reasonable accuracy elsewhere on the curve.
So what did I discover ? Well the camera is definitely limited in terms of the temperature range that it can display on its built in screen. It is apparently unable to show a temperature below 25C or above around 50C. You can save a radiometric JPG image to the micro SD card and analyse it using the free Hikvision image analysis software. Now this software is not going to set the World on fire with its amazing capabilities, but first impressions are not bad for such a free software. The user directs the software to the location of the images and then selects the image of interest. The thermal image appears in the work area and may be further anaysed with simple tools such as a temperature line, circle, oval polygon or spot to show details of the pixel temperatures in a Region of interest. The images are fully radiometric. When an image from a DS-2TP31B-3AUF is loaded into the software, the image appears in the work area with a span automatically set by the software to cover the minimum and maximum tempertures present in the scene. There is no 25C to 50C limitation on the span :-+ From the sample supplied images I was able to determine that the Fever Detection camera actually stores a fully radiometric image that covers the normal -20C to +150C capability of range 1 on a DS-2TP31 standard model camera. I can set the Auto Span mode to manual and change the minimum and maximum displayed scene temperature with ease in the range -20C to +150C. The Tea test was showing temperatures in excess of the 50C upper limit of the Fever Camera with no warnings or obvious issues. Was the reading accurate though ? Hmmm on that I will say I have no proof either way :-// Is the high temperature range operational ? That range covers 0C to 550C but you cannot select that range on the camera so I suspect the answer is no. I was able to select a palette from a large list of available LUT's and was pleased that I also had the option to switch off the centre, highest and lowest temperature cursors if I wanted a 'clean' image. The analysis software allows the user to change palettes and other settings so it is actually a useful tool and indispensible if trying to use the Fever Detection camera outside of its normal Fever Detection temperature measurement range. Note that the "live Image" option offered at the top of the software is only detailed in the manual for the DS-2TP21 model of camera so the DS-2TP31 may not provide the required data streaming functionality :(
Finally, to image quality ....... Hmmm something seems a little odd in this area of performance. Initial impressions of the supplied images were not good. Curved edges look stepped and the whole image seemed more 'granular' than I am used to on a 160 x 120 pixel image... but not noisy :-+. It is as though no attempt is being made to soften edges and disguise the pixelation present in a low resolution image. This is likely very deliberate and necessary. This camera is being used as a radiometric measurement device and heavy image processing can negatively impact temperature measurement accuracy. As I often say.... in thermography, it is all about clean, accurate temperature data and not about pretty pictures ! I believe that the Fever Detection camera is only applying the minimum required image processing in order to preserve the thermal data. There are multiple frames captured in the design of the measurement system and this may also add to the somewhat unusual imaging that I am seeing.
At the end of the day, this is a 160 x 120 pixel temperature measurement tool. It may well outperform general use thermal imagers in terms of accuracy as a result of the thought that has gone into making the camera as accurate as possible from 30C to 45C.
Lots to think about here. Such a pity that the Fever Detection mode cannot be switched off to enable general use but Hikvision were clearly aiming at the Fever Detection market and likley wanted to ensure that the user could not easily mess up measurements, as could occur if the camera was reverted to a 'pretty pictures' mode. Such an approach also exists in Fire Fighting cameras where the KISS principle is practiced when designing the user controls and interface.
I attach a sample PDF report that I created with the supplied Teacup image. I also attach a screen capture of the software with a Hikvision supplied image that is is from a different camera model.
Fraser
-
Some sample images from a DS-2TP31B-AUF camera for you to study :)
These will load into the Hikvision software if you want to play with it. These are pictures of a Tea cup/mug, Gas hob flame and contents of a fridge.
Software is to be found on the Hikvision site...here....
https://www.hikvision.com/europe/support/download/software/ivms-4800-temperature-measurement-/ (https://www.hikvision.com/europe/support/download/software/ivms-4800-temperature-measurement-/)
Fraser
-
Screen grabs from the software whilst playing with the sample images :)
After seeing these images, is the camera worth £60 to £70 ? I think it is but measurement accuracy remains an unknown that I cannot be certain about unless I test a camera against my blackbody sources.
Enjoy :-+
-
So the thermal imager works perfectly normally, scaling the colours as usual and the only 'fever detection' aspect is the on-screen temperature text. Seems very odd having gone to the other lengths Fraser has described yet not to cap and collar the colours
So down to the code folks to find the four lines like
'If Tmin <25 then Tmin(text) = "<25" '
bodged into the on-screen display code and change them back
-
From what I can tell, the cameras range selection menu is locked to the high sensitivity -20C to 150C range and cannot be changed to the low sensitivity 0C to 550C range. Again a simple change to the code to disable the range selection menu entry and change the menu text from “-20C to 150C” to “30C to 45C”. We must not forget the other changes though. Apparently the camera needs to be held static for best performance and I understand that multiple frames are stacked in the measurement process. It is not super-resolution as that requires movement to be effective. There may also be changes to the calibration regime as Hikvision are claiming +/-0.5C measurement accuracy over 30C to 45C. From what I can see the DS-2TP31B-3AUF has been tuned to the intended role and the displayed information is tuned to the fever detection role, including the palette scale. The end stops for the palette LUT appear to be 25C to 50C. The Radiometric data recorded on saved images appears to be ‘raw’ and treated no differently to a standard DS-2TP31 version of the camera when read by the image analysis software. Sadly that is not the most convenient of setups for many users however. The software only interprets and displays the raw data with which it is provided so sadly it cannot control, the camera or do anything to enable the low sensitivity range of 0 to 500C as that is set up in the camera
This has been an interesting investigation into the Hikvision Fever Detection camera but highlights the complications of buying Fever Detection cameras with the intention of using them for general applications or PCB work. People have asked me whether I think there will be masses of Fever Detection cameras appearing as surplus on eBay and at auctions, once the Pandemic is declared ‘Over’. In truth I have no idea but this eBay auction suggests that, on a small scale , this may already be happening. Thermal imaging cameras are not able to show a fever in a person who is not suffering one and we know that Corona Virus is infectious before a Fever Occurs ! Confidence in Fever Detection cameras may decrease, leading to disposal of new and used units. Be warned that you need to consider what I have detailed in this thread if you intend to use a Fever Detection camera in a role that is incompatible with the needs of Fever Detection.
I hope this thread helps readers to buy from a position of knowledge and avoid mistakes.
Fraser
-
Also…. Be warned that there are sellers on eBay offering the DS-2TP31B-3AUF as the full -20C to 550C measurement range DS-2TP31 model because the included user manual in the box is the wrong one for the camera ! I have just finished helping a seller to identify the actual camera that they are selling as they were very confused by the conflicting camera model number and specifications shown in the user manual. We proved that the manual was the wrong one. Watch out for that all important “B” in the cameras part number !
Fraser
-
From what I can see the DS-2TP31B-3AUF has been tuned to the intended role and the displayed information is tuned to the fever detection role, including the palette scale. The end stops for the palette LUT appear to be 25C to 50C.
The image of your fridge seemed to show your fridge (5°C) at mid-range. Is that a PC-software image rather than a screenshot as I had assumed ?
-
Now you have got me thinking Bill !
Having revisited the sample images I thought I would add a commentary to what we are seeing. I attach both the image from the cameras memory that may be viewed as a standard JPG image and also the image as presented on the PC image analysis software. I added some ROI measurement points to capture temperature data around the scene.
OK, thinking out loud here…..
The camera may be considered a standards DS-2TP31 in terms of the palettes and how they are applied to a thermal scene.
The Maximum temperature present in the scene should rack across to the top colour of the Palette LUT
The Minimum temperature present in the scene should track across to the Tom colour of the Palette LUT
This is normal auto span behaviour and ensures that the Palette LUT is applied to best effect in the scene display. Nothing unusual here then. However, once the scene contains upper or lower temperatures that extend beyond the range 25C to 50C, the temperature markers on the Palette Bar change to show that the upper or lower measurement limit has been exceeded (<25C or >50C shown)
In the case of the fridge image, all parts of the scene are correctly imaged according to the palette side bar but as they are all below the 25C lower measurement limit, the Palette bar has the <25C indicator at both ends of its scale so you cannot see the span limits set by the cameras auto ranging function. This looks weird but does make sense in a Fever Detection Camera . If that same image is saved to the flash memory card, all Radiometric data is saved but no Fever Detection firmware limits are applied or included in the metadata package. As such the PC image analysis software presents the full image information, including true upper and lower settings of the span and associated palette bar limits. On the right hand panel there is a range selection box that offers -20C to 150C and 0C to 550C ranges but there is nonFever Detection range of 25C to 50C present. Note that the 0C to 550C range does not enable use of the Fever Detection Camera at temperatures above 150C as the camera has to provide that higher range with appropriate calibration tables etc.
So on purely the visual experience, the Fever Detection Camera effectively declines to show span edges beyond 25C to 50C yet behind the scenes the Radiometric data is still available for post capture analysis. There is still the matter of the Measurement Calibration, that may differ from the standard version of the camera, and the multi frame capture used in the Fever Detection version.
Fraser
-
I may be able to buy one of these cameras for a teardown after Christmas, funds permitting. I advise caution if considering one of these until I have carried out a more detailed study of the hardware design and firmware. The UNI-T UTi120S, UTi690A, UTi260B or Infiray C210 may be a better option for many users as these offer full measurement capabilities ‘out of the box’. A bargain, is only a bargain if it does what you need of it ! The DS-2TP31B-3AUF appears to be a somewhat unusual camera in the current market due to its need to be accurate over a narrow temperature range for Fever Detection. This specialist need has required changes in firmware that are not well suited to general use. For this reason, I currently cannot recommend it, especially now that Uni-T and Infiray are offering very reasonably priced general use cameras and phone dongles.
In précis…known issues with the DS-2TP32B-3AUF Fever detection camera…..
1. The temperature measurement capability on the camera is limited to the range 25C to 50C with only the range from 30C to 45C stated as calibrated for a +/-0.5C tolerance. Measurement accuracy outside this stated range is not known at this time and could potentially be poor.
2. The camera employs a multi frame measurement system that requires the camera to be static on a tripod rather than hand held if accurate measurements are required. This frame accumulation process for measurement appears to cause issues in the image if the camera is panned, evenly slowly. This effect needs to be investigated further.
3. The camera has its low sensitivity (0 to 550C) measurement range disabled in firmware. This range may, or may not, have an associated calibration table and if such is missing from this camera build, no amount of firmware tweaking will recover this mode in a useful form. This range is desirable in electronics where temperatures exceeding the High Sensitivity ranges 150C upper limit are possible.
4. There is no option to switch off the high temperature tracking cursor on this camera and that can become an annoyance to some users. The centre temperature cursor may be switched off however. A clean image without cursors is possible through use of the PC based analysis software, if desired, but the saved jpg images on the cameras memory card do contain whatever cursors were displayed at the time of saving.
5. The Fever Detection version of this camera uses a special firmware that is available to download. Unfortunately, and surprisingly, the firmware for the standard -20C to 550C capable model is not available. This may be because no updated firmware has been released for the standard model since initial release. This does mean that there is no replacement firmware available to modify the Fever Detection Camera to a standard ‘General Use’ Firmware build :( I have not found a similar Hikvision camera firmware from which to take any elements that could help in the conversion of the Fever Detection firmware build.
More on this camera if I obtain one for further investigation. I am subconsciously talking myself out of that idea as I type this post though :-DD
Fraser
-
Hello Fraser and Happy new year.
What an interesting coincidence, I acquired one of these cameras exactly a year ago, for 100€, thinking I had the deal of the century given the 160x120 resolution @ 25Hz, and a 999.999€ price-tag.
My intend was to check if such device could be turned into a homemade cheap equivalent of Flir Scout-TK, to observe wildlife in my area, without any measurements considerations. Its very wide field of view, 50x37º was definitely a deal-maker.
Upon unboxing, I was disappointed by the level of noise in the live stream, but still, pictures were more than decent for a so-called «fever detector».
From that point, I've been waiting for some thermography magician to open it up, discover a video output, or better, some code able to stream through USB...
If Santa Claus brought you a unit to disassemble, be sure you'll have a reader to check this thread.
Cheers