As someone who has ordered a SEEK unit, I should state that I am NOT expecting the imagery to be as good at that which I get from my other thermal cameras. Why? Well this is a new camera from a company that AFAIK has not written thermal camera image processing software before (unconfirmed) They may have an excellent microbolometer from Raytheon but as seasoned TIC designers will tell you..... there is much to be done in terms of image processing to make a decent image. SEEK are likely to refine their image processing software over time (just like FLIR did) and we should not expect too much in the first 'cut' of public release software.
The microbolometer is a compromise device....it always was and likely always will be. The compromise is in noise figure. The modern cameras are far less noisy than the early ones like my FLIR PM570. Ever since FLIR released the PM570 in 1997 they have been working on reducing the noise content of the images. Some has been through advances in core technology whilst much has been achieved with advanced image processing algorithms and filtering.
I remember when I first tested the PM570. I was one of the first in the UK to do so. I did not like its images at all and dismissed it as too noisy. Even FLIR admitted that the image noise content was an issue and this would be addressed in later firmware releases. Compared to what you may ask..... well compared to a FLIR PM550 Stirling cycle mechanically cooled camera. The PM550 looked like the later PM series cameras but it is the daddy of them all. The thermal imaging array was the same resolution as the PM570 at 320x240, but it was semiconductor based, cooled to -196 C (-371 F or 77K). At such a low temperature the sensor pixels and associated components within the FPA are extremely low noise. The images produced by that camera have none of the noise that we are used to with Microbolometers, even when set to a 2 C span. They are truly things of beauty. The uncooled Microbolometer based cameras will have a very hard job to equal the performance of such a cooled sensor as the law of physics is against such. For low noise you really want cooling and not room temperature. Just look at the technology used in the Low Noise Amplifiers of Radio Telescopes for more proof of this (they are cooled to the same temperature as the PM550 sensors array).
Back to the SEEK TIC. This new core will likely have been built within a tight budget and so we should expect it to be a little noisy as it may not have the clever tricks of higher end microbolometers included. The lens size is a handicap due to small size and material used. This is another compromise that has had to be made in order to reduce cost. As stated, much will also depend upon the post capture processing carried out in the SEEK camera module and host computer (phone). It is reasonably straight forward to make a good looking image by filtering and softening of edges. That is not to say that such an image is good for radiometric usage though....it just looks pretty but some detail is lost as a consequence. If you wish to see a cameras true self, set it to the smallest span possible, usually 2 C and watch the image noise appear ! At such small spans its is very hard to suppress the noise content. Many manufacturers do not offer very small temperature spans as a result. This isn't fair criticism in this case though. $200 for the SEEK is an amazing deal IMHO which is why I bought one and my colleagues are dying to get their hands on it ! I see a bright future for companies like SEEK and they will hopefully keep improving their product and the software that processes the images. Its early days after all.
Which would I prefer.... the FLIR ONE with its fancy case or the mores agricultural looking SEEK ? No contest, the SEEK has far more potential with its higher resolution core and a potentially greater interface capability for Hosts. It does not need MSX to do a decent job. The FLIR PM series are respected throughout industry as high performing, reliable thermal cameras...... and none had MSX
It all comes down to intelligent usage and a trained operator
MSX is a nice function to have though..... no argument there.
Aurora
Home
Help
Search
Login
Register
