New Flir products

[Ben321]

Looks way too complex, just like official FLIR software like FLIR Tools. I just need a simple program that extracts the 16-bits-per-pixel array from the FLIR proprietary image and saves it into a TIFF file (TIFF does indeed support 16-bits grayscale images). Then I could load it into my own program for further processing. I'm not particularly interested in the actual temperature readings, so the emissivity and stuff does NOT need to be saved for my application. It can all be discarded except for the actual 16-bits-per-pixel array, which should be saved into a raw format that supports 16-bits grayscale images. And the first format that comes to mind that supports that is TIFF. That program you mentioned to me, as you can see is actually over-bloated for my purposes, handling all that other thermal info (like the emissivity, Boltzmann constants, etc). I don't need any of that, just a way to visualize the raw pixels as created by the far infrared focal-plane-array in the camera.

I fear that such software doesn't exist, but if it does, my best chance would be asking on this forum. I can see my worst fear is true. Software that meets my above mentioned requirements does not exist. Both pieces of software you mentioned are overbloated, and therefore do not meet my requirements.

My requirement is that the software I'm looking for do the things I mentioned, and nothing else, in order to make it a relatively slim software package. It should have the feel of a bare-bones simple image viewer, not a professional thermography program.

[Ben321]

160x120 pixels confirmed. Thermal image is upscaled.

Aurora

Do you know of any FREE software that can extract thermal image raw data from FLIR radiometric images?

ExifTool by Phil Harvey http://www.sno.phy.queensu.ca/~phil/exiftool/ is what has been used to extract data in the past. It's command line, very powerful but be ready to do some reading!

I was hoping you could point me to a dedicated FFF reader (not something like ExifTool, which has all kinds of uses). Preferably a GUI based software created by a 3rd party programmer, who had previously reverse engineered Flir's FFF file format.

Perhaps try this?

I saw it mentioned in the FLIR E4 thread.

Ok, so I finally completed my FLIR image parser that lets me extract the raw thermal pixel from FLIR radiometric files now. Now all I need to do is make an application that uses this framework, and I'll have what I'm looking for.

[tomas123]

Do you know of any FREE software that can extract thermal image raw data from FLIR radiometric images?

I was hoping you could point me to a dedicated FFF reader (not something like ExifTool, which has all kinds of uses). Preferably a GUI based software created by a 3rd party programmer, who had previously reverse engineered Flir's FFF file format.

I don't understand your question.
Exiftool is a simple tool for extracting the raw data of multiple images in one line:

Code: [Select]

// Windows Code
exiftool -b -RawThermalImage FLIR*.jpg -w _%%f.png
but it's a long way from RAW values to temperature 

see source code for extracting the FFF header:
http://cpansearch.perl.org/src/EXIFTOOL/Image-ExifTool-9.90/lib/Image/ExifTool/FLIR.pm

see attachment for further details

a dedicated FFF reader (not something like ExifTool, which has all kinds of uses)

two years ago I wrote a simple awk script for extract the raw values from jpg:
http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,4898.msg23497.html#msg23497


PS: Flir Tools is a "free" GUI Software...

[Ben321]

Do you know of any FREE software that can extract thermal image raw data from FLIR radiometric images?

I was hoping you could point me to a dedicated FFF reader (not something like ExifTool, which has all kinds of uses). Preferably a GUI based software created by a 3rd party programmer, who had previously reverse engineered Flir's FFF file format.

I don't understand your question.
Exiftool is a simple tool for extracting the raw data of multiple images in one line:

Code: [Select]

// Windows Code
exiftool -b -RawThermalImage FLIR*.jpg -w _%%f.png
but it's a long way from RAW values to temperature 

see source code for extracting the FFF header:
http://cpansearch.perl.org/src/EXIFTOOL/Image-ExifTool-9.90/lib/Image/ExifTool/FLIR.pm

see attachment for further details

a dedicated FFF reader (not something like ExifTool, which has all kinds of uses)

two years ago I wrote a simple awk script for extract the raw values from jpg:
http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,4898.msg23497.html#msg23497


PS: Flir Tools is a "free" GUI Software...

Problem with the of the PNG FLIR images is that the 16-bits-per-pixel data that is decompressed from them, you'll find out very quickly is little-endian. Yet the PNG specification states that all multi-byte values (including pixel values if the bit-depth is greater than 8-bits-per-pixel), must be big-endian. So no PNG viewer that meets the official specs will be able to properly decode the image. I wish that exiftool had a way to swap the byte-order (endianness) of the decompressed pixels from the PNG image. Even better yet, I wish it had a way to convert the embedded PNG image to TIFF, so that the pixel values were directly accessible without additional decompression steps by the next program in the tool chain (an image viewer).

Fortunately this is all unnecessary now, as I have completed my FLIR image parser component, that detects any form of FLIR image (whether in a standalone FFF file, or embedded in a JPEG; and whether raw, or PNG compressed) and takes whatever steps necessary to convert get a raw 2D array of 2-byte-per-pixel values in little-endian byte order (even swapping the byte order if needed to accomplish this).  The output array is a in the form of signed 32bit Long integers (not unsigned 16bit Short integers), as Visual Basic 6 (the programming language I used for creating this component) only handles signed numbers. It is 100% incapable of handling unsigned numbers. So each value in the output array always has the upper 2 bytes at 0 (zero), and the lower 2 bytes contain the pixel value. Why use VB6? It's the easiest BY FAR programming language to use. If I tried to do the same thing in Visual C++, I'd still be stuck trying to figure out how to accomplish all the things that this component needs to do. I plan to release this component at some point as an ActiveX DLL file for other developers to use in their projects, but first I'll be using it as the core of the application that I'm writing (also in VB6), which will be a very basic FLIR image viewer. I will likely release these 2 things (both the application, and the ActiveX DLL file) at the same time.

[tomas123]

Do you know of any FREE software that can extract thermal image raw data from FLIR radiometric images?

see topic 4 "Tutorial: create images with exiftool and Imagemagick step-by-step" of my post
https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg342072/#msg342072

Why, do you don't read my post?

The nice guys from imagemagick have extended the software for Flir to swap the endian order in PNG images:
http://www.imagemagick.org/discourse-server/viewtopic.php?t=24818

// swap byte order
since IM 6.8.8 with switch png:swap-bytes

Code: [Select]

convert -define png:swap-bytes=on t1.png t2.png

I have completed my FLIR image parser component
...
I will likely release these 2 things (both the application, and the ActiveX DLL file) at the same time.

is the code open source?

[Fraser]

@Bud,

Nice picture.

It also demonstrates the challenges of using a simple thermal camera in an outside environemnet

Note the temperature span that the camera has selected.....-40C to +26C. A 66C span 


It has done this because the image contains the sky and so the autospan tries to include all temperature present in the FOV. (-40C is as low as she can go) A wide span reduces contrast in the image and it becomes more difficult to see subtle or small temperature differences. The large span effectively flattens the image.

If the same image were to be produced using a manual span setting of say 20C to 30C or 10C to 30C you would see a very different image  Simple thermal cameras are great for 'point and shoot' applications but there is a need for a manual span and centre temperature control to be used if the best is to be be extracted form the image. Unlike a full-auto visual spectrum digital camera, Auto mode in a thermal camera often produces images that are far from optimal for thermographic work....they just produce 'pretty' pictures. The good news is that many thermal cameras have post capture image processing software for a PC. This can change the span and centre temperature, completely changing the image produced. This is the huge advantage of a radiometric image format as opposed to a simple non radiometric file type. In a radiometric image, all the data from the imaging chip is present (the full temperature range) in the file and this may be manipulated as the user desires.

If you place your cat picture into FLIR tools you should be able to change the span and centre temperature to show the cat more clearly against its background.

It is the flexibility of radiometric image files that makes them so wonderful for pulling out 'hidden' details that might otherwise be missed and would be lost in a non radiometric file format.

Aurora

[tomas123]

nice samples
http://www.robotroom.com/Flir-Infrared-Camera-3.html

[Fraser]

@tomas123

 

Aurora

[joe-c]

can somebody provide some images from the flir one 2 while view long range?
direct images (radiometric) would be nice. 

[encryptededdy]

can somebody provide some images from the flir one 2 while view long range?
direct images (radiometric) would be nice. 

http://www.ar15.com/forums/t_6_18/448361_2nd_Gen_FLIR_One.html&page=1#i4371588

Unfortunately not radiometric JPG.

[Fraser]

Ref the images of two horses in a paddock.......

As already detailed above, such exterior images that include the sky will likely force the Auto span of the FLIR ONE to go wider than optimum for spotting small animals with a challenging thermal profile. The relatively wide field of view provided by the FLIR ONE's lens also makes it sub-optimal for small target detection at long ranges.

It isn't doing too badly with the horses, but then they are very large and hot targets 

I have attached the images to this post in case they are lost in the AR15.com site. These are attributed to 'Skypup' on that forum.

The distance to the hoses is estimated to be 250m (75 metres).

The AR15 site is one to keep an eye on if interested in longer range thermal cameras as they have many members who use the thermal domain for the detection of wild life. It has some interesting 'real world' experiences of such and often pictures to support such. I am not a hunter myself, but those guys are certainly embracing the newer, more affordable, thermal camera technology in a way very different to the OEM's original intention. The SEEK camera is also being used by these guys.

Aurora

[Ben321]

I have completed my FLIR image parser component
...
I will likely release these 2 things (both the application, and the ActiveX DLL file) at the same time.

is the code open source?

Well since I finished my program before I finished the component, I'm releasing the program now. The standalone program has all the features that the component will have, except that it's compiled right into the exe file, instead of the program having the separate component as an added dependency. My next plan is to create the component, which will be an ActiveX DLL file. It should be a fairly simple process of migrating this code to a new VB6 project (the one for generating this component) now that I have all the code I want laid out in nice *.bas VB6 module files. I hope to complete the component, a single ActiveX DLL file, some time tomorrow, so you guys will be able to easily implement this functionality in your own programs, without having a bunch of module files that you would need to add to each project to get the same functionality.

Please read the readme file in the zip file, so you will know how to get the program setup to run. It's a very simple program to use. It's very self explanatory how to use each of the functions. So I haven't put a help file in the zip file.

The program, which is called FLIR FFF Viewer, is able to load all of the forms in which FLIR FFF files can exist, including ones with PNG compressed raw data, and the ones in which the FFF file is not a standalone file, but rather is embedded in a JPG file. Once loaded, there are a number of options for how to display it. It offers the ability to save the image as it is currently displayed to a BMP file. It also offers the ability to take the raw 16bits-per-pixel thermal image and save it to either an uncompressed TIF file or a losslessly compressed PNG file. Unlike with FLIR's implementation, which saves the 16bit raw data in little-endian byte order for PNG files (even though multibyte image data in PNG files is supposed to be big-endian), my implementation of 16bits PNG saving actually correctly swaps the bytes to save big-endian image data in the PNG file.

FFF files can be in one of 4 formats (all 4 of which my program will handle). FLIR calls these:
FFF
FFF+PNG
JPEG+FFF
JPEG+PNG

FFF and FFF+PNG are standalone *.FFF files. FFF stores the 16bit image data uncompressed (either in little-endian or big-endian byte order), while FFF+PNG stores the data as a losslessly compressed PNG file embedded within the FFF file (though the uncompressed image data that is input to the compressor is in the wrong byte order for PNG, thus it is technically an invalid PNG file).
JPEG+FFF and JPEG+PNG are *.JPG files with embedded *.FFF files. For the JPEG+FFF format, the embedded *.FFF file is in the pure FFF format. For the JPEG+PNG format, the embedded *.FFF file is in the FFF+PNG format.

For JPEG+FFF and JPEG+PNG, the embedded FFF file is extracted from the JPEG container prior to further processing. With the extracted (or standalone) FFF file, if it is of the FFF format, my program will directly extract the raw data. If it is in little-endian byte order this pixel array is passed directly to the rest of the program. If it is in big-endian byte order the bytes are swapped to little-endian (as needed in Windows software) before it is passed to the rest of the program. If the extracted (or standalone) FFF file is of the FFF+PNG format, my program uses ZLib (in the form of zlibwapi.dll, the stdcall version of the DLL file) to decompress the image data in the PNG file, and then passes the decompressed pixel array to the rest of my program. This is the required core functionality of any program that is to handle FLIR radiometric images.

All this functionality will also be wrapped into my ActiveX DLL component when I finish making it, allowing others to write their own programs for processing FLIR radiometric images.


Here's the download link for my program (not the DLL component). And yes, it does have all the source code.
Version 1.0 https://www.mediafire.com/?9n5m7hstbats5n9
Version 1.1 https://www.mediafire.com/?o5wjeonnc390jco
Version 1.2 https://www.mediafire.com/?zf2imwze56kqvc4

[tomas123]

 
quick check on Windows 8.1 x64

(1) need zlibwapi.dll from folder "Other Dependencies" ( Error message: MsgBox "Failed to load image. File is not a valid FLIR radiometric image.")

(2) there are many of my Flir images, which your tool can't read ("Failed to load image. File is not a valid FLIR radiometric image.")

(2) you can test it with the nice Flir image collection of Phil (52 Flir cameras!)
http://owl.phy.queensu.ca/~phil/exiftool/sample_images.html
http://owl.phy.queensu.ca/~phil/exiftool/FLIR.tar.gz

[Ben321]

quick check on Windows 8.1 x64

(1) need zlibwapi.dll from folder "Other Dependencies" ( Error message: MsgBox "Failed to load image. File is not a valid FLIR radiometric image.")

(2) there are many of my Flir images, which your tool can't read ("Failed to load image. File is not a valid FLIR radiometric image.")

(2) you can test it with the nice Flir image collection of Phil (52 Flir cameras!)
http://owl.phy.queensu.ca/~phil/exiftool/sample_images.html
http://owl.phy.queensu.ca/~phil/exiftool/FLIR.tar.gz

Can you please send me the specific files that you had trouble with? I would like to test those, and see where my program is having problems.

[tomas123]

use this images

(3) you can test it with the nice Flir image collection of Phil (52 Flir cameras!)
http://owl.phy.queensu.ca/~phil/exiftool/sample_images.html
http://owl.phy.queensu.ca/~phil/exiftool/FLIR.tar.gz

[Ben321]

use this images

(3) you can test it with the nice Flir image collection of Phil (52 Flir cameras!)
http://owl.phy.queensu.ca/~phil/exiftool/sample_images.html
http://owl.phy.queensu.ca/~phil/exiftool/FLIR.tar.gz

I don't want to have to sort through 52 photos to find a few that happen to for some reason don't work with my program. So far it's worked with other FLIR radiometric images I've tried, that I've found via Google Images search. Not one has failed yet. So I think it's probably a very rare type that isn't working. That's why I'm asking that you send me the exact files that aren't working for you. Upload the ones you tried that it isn't working with to Mediafire (or other similar file sharing site) and then post the link here for me to download them.

Update:
I just downloaded the 52 picture set you recommended, and every single picture in that set is corrupt. If that's the set that you got your picture from that you claimed didn't work with my tool (rather than the picture you tested having come from your own thermal imager, which is where I have been assuming it came from) then it's no wonder the picture you tested didn't work. Those pictures aren't even valid JPEG files. All you get when you try to view those images distributed in this massive collection of 52 corrupt JPEGs, is an 8x8 white square. I have never seen any picture file so corrupt before, let alone an entire collection of 52 corrupt images. Looks like if I want a picture from each model of thermal imager that FLIR has ever manufactured, I'm gonna have to ask some employee at FLIR who has access to all the cameras to take some pics for me, and email them to me.

[Ben321]

Update2 part 1:
Ok, I see why those 52 JPEG FLIR images don't show up. The main image in the JPEG file was replaced with an 8x8 blank image, but the rest of the metadata remains, including the embedded radiometric pixel data. Note that the image called FLIR_ThermaCAM_A320.jpg is missing radiometric pixel data, and so it has nothing to be displayed in my program. Therefore my program will always report it as in invalid FLIR radiometric image.

Update2 part 2:
I have now completed fixing the bugs in my program. All 52 of those sample images (other than the one from the Thermacam A320) will now load in my program. The previous version with the bugs in question was version 1.0. This current version with the bug fixes is version 1.1. Download it here: https://www.mediafire.com/?o5wjeonnc390jco

I've also put this link in the original post where I first posted version 1.0.

[encryptededdy]

Update2 part 1:
Ok, I see why those 52 JPEG FLIR images don't show up. The main image in the JPEG file was replaced with an 8x8 blank image, but the rest of the metadata remains, including the embedded radiometric pixel data. Note that the image called FLIR_ThermaCAM_A320.jpg is missing radiometric pixel data, and so it has nothing to be displayed in my program. Therefore my program will always report it as in invalid FLIR radiometric image.

Update2 part 2:
I have now completed fixing the bugs in my program. All 52 of those sample images (other than the one from the Thermacam A320) will now load in my program. The previous version with the bugs in question was version 1.0. This current version with the bug fixes is version 1.1. Download it here: https://www.mediafire.com/?o5wjeonnc390jco

I've also put this link in the original post where I first posted version 1.0.

Oddly I can load FLIR One pictures fine, but once FLIR Tools have touched them they are no longer openable. I assume they're still radiometric JPEGs (I could be wrong) as FLIR Tools is still happy to open them. I've attached a copy of a image before and after I edited it with FLIR Tools.

[tomas123]

@Ben321
with old Version Flir Tools V 4.0 it works fine (look at attachment)

see here:
http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,6119.0.html

it looks like newer versions of FLIR Tools has changed the way EXIF data is being embedded in images it saves.
(FLIR Tools 5.0 (5.0.14283.1002))
...
The problem is that they changed the version number (from 100 to 101):

exiftool source code:
http://cpansearch.perl.org/src/EXIFTOOL/Image-ExifTool-9.90/lib/Image/ExifTool/FLIR.pm

   # FLIR file header (ref 3)
    # 0x00 - string[4] file format ID = "FFF\0"
    # 0x04 - string[16] file creator: seen "\0","MTX IR\0","CAMCTRL\0"
    # 0x14 - int32u file format version = 100 /101
    # 0x18 - int32u offset to record directory
    # 0x1c - int32u number of entries in record directory
    # 0x20 - int32u next free index ID = 2
    # 0x24 - int16u swap pattern = 0 (?)
    # 0x28 - int16u[7] spares
    # 0x34 - int32u[2] reserved
    # 0x3c - int32u checksum

    # determine byte ordering by validating version number
    # (in my samples FLIR APP1 is big-endian, FFF files are little-endian)
    for ($i=0; ; ++$i) {
        my $ver = Get32u(\$hdr, 0x14);
        last if $ver >= 100 and $ver < 200; # (have seen 100 and 101 - PH)
        ToggleByteOrder();
        $i and $et->Warn("Unsupported FLIR $type version"), return 1;
    }

your source code

FffVersion = CopyLong(FffFile(&H14))
Select Case FffVersion
    Case &H64&
        BigEndian = False
    Case &H64000000   ' = 100
        BigEndian = True
    Case Else
        ReDim ThermalPixels(0, 0)
        ThermalPixels(0, 0) = -1
        GoTo EndOfFunction
End Select

[Ben321]

Oddly I can load FLIR One pictures fine, but once FLIR Tools have touched them they are no longer openable. I assume they're still radiometric JPEGs (I could be wrong) as FLIR Tools is still happy to open them. I've attached a copy of a image before and after I edited it with FLIR Tools.

Yep, I just found out why. As of FLIR Tools version 5, there is a new version number for FFF files. The version number is stored in the FFF file as a 4-byte dword value. It normally is represented by the hex value 0x00000064, which is decimal value 100. Divide the value stored in this field by 100 to get the actual version number. The previous version field value has been 0x00000064 which represents FFF file version version 1.00. However as of FLIR Tools version 5, the FFF version field number is 0x00000065, which represents FFF file version 1.01. I will create a new version of my FLIR FFF Viewer application to take this into account. Previously it used this version field to determine the endian-ness of the main header, as there is no explicit "endian field" in the header to allow a program to detect the endian-ness of the main header of the FFF file. That's why I had used the "version field" in the header to detect the endian-ness. If it is little endian FFF file header then the hex bytes in this field would be "64 00 00 00". If it is big endian, then the hex bytes in this field would be "00 00 00 64". These 2 specific byte patterns are explicitly written into my program to allow the endian-ness to be detected. Any other values it is programmed to assume are from an invalid FFF file. With the new version of FFF file that comes with the latest version of FLIR Tools though, I will now need to change how my program detects the endian-ness of the main header of the FFF file. If anybody here has a suggestion on how to do that, or if there's a secret undocumented field hidden somewhere in the FFF file that specifies the endian-ness of the main header, I hope somebody here can inform me of it. For now, I'm going to use a sort of "hack" on my version field detection of endian-ness, that should (at least for the time) work properly with all current known versions of the FFF file format.

[Ben321]

Ok. Here's the latest version of my program (v1.2). It is now able to work with all versions of FFF file, including those generated by the latest version of FLIR Tools.
https://www.mediafire.com/?zf2imwze56kqvc4

I have also posted this link now back in my first post that mentions my software.

[tomas123]

The previous version field value has been 0x00000064 which represents FFF file version version 1.00. However as of FLIR Tools version 5, the FFF version field number is 0x00000065, which represents FFF file version 1.01.

Yes, but do you haven't read my earlier post?

Code: [Select]

FffVersion = CopyLong(FffFile(&H14))
If (FffVersion And &HFFFF&) = 0 Then BigEndian = True Else BigEndian = False


[Ben321]

The previous version field value has been 0x00000064 which represents FFF file version version 1.00. However as of FLIR Tools version 5, the FFF version field number is 0x00000065, which represents FFF file version 1.01.

Yes, but do you haven't read my earlier post?

Code: [Select]

FffVersion = CopyLong(FffFile(&H14))
If (FffVersion And &HFFFF&) = 0 Then BigEndian = True Else BigEndian = False


What's wrong with it now? I got it working on all known FFF files. There is no FFF file that it will not work with now. So I don't know what you think is wrong with it. It would be helpful instead of just quoting my post and then saying that I should have read an earlier post of yours, if you actually said "this is what's wrong with it...". Otherwise it seems like you are just trying to find criticism with my program, without providing any constructive information on how to improve it.

Well it works now, so unless you can point out an FFF file that it doesn't work with, I'm through with altering the program, and my most recent version (v1.2) will remain the final version of this program. You can use it if you want, or don't use it if you don't want. As far as I'm concerned, this program is vastly superior to FLIR Tools for what I need it for. I don't need 100 different ways to perform temperature measurements. I just need a way to view these images, save the image as viewed for a reasonable range of brightness levels, and also convert the raw data to more common image formats. My program does all of these things, without having as many overbloated extra features as the official FLIR Tools program or the unofficial BFIC program, and also without being such a huge program (filesize wise), unlike FLIR Tools which is hundreds of megabytes in size. My program I consider the ideal program, that everybody who handles thermal images should use as a good starting program (and they should only turn to the other programs if they need the extra features). So as I said, you can use it if you want or don't use it if you don't want, but unless there is a VERY urgent bug remaining in it (one which prevents it from loading certain FFF files) that you can point out, then I have no plans to make any further "corrections". It is entirely possible that some of the code is not "technically" the right code for accomplishing what it is designed to accomplish, but as long as it works, I have no plans to change it.

Rather than work on my standalone application anymore, I am going to spend my time porting the loader part of my application to its own separate component (an ActiveX DLL file), so that others can incorporate it into their own programs.

[Ben321]

I've completed the porting of the loader code from my application over to my ActiveX DLL component. You can download it here.
https://www.mediafire.com/?7wo4m2m0b45u1bm

Please take care to read the readme file for this, as there's a lot more complexity when it comes to using ActiveX DLL files than any other kind of binary/executable file (such as EXE files or standard DLL files).

[tomas123]

I just wanted to help you.
With Phil (exiftool) together we reversed engineered two years ago the data structure of flir radiometric jpg and the formulas for calculating temperature from RAW values.
We published our findings and I'm glad when these things will be used from the community.