Wow, what a topic to talk about! As these words as mentioned above, with very different meanings, are spoken daily by people in all walks of life, from children to adults, from school areas to the workplace and even around the supper table, if not in front of the telly.
No, you are wrong, I can see what you think and it is not about that “steaming” topic. I would like to talk about something completely different, but with the same headline sentence. No age restriction at all! Is it possible? Yes, read for yourself!
Today, there are many thermal camera types and the most general and most commonly used criteria are the so called “quality” of the image! How many times have I come across a sale where the customers have said, “Look how beautiful is this image, compared to…..” But, what does “Quality of the image” mean, is it the nice image, is it the crisp and clear image, and is it a function of the camera or analytical software or the monitor even?
Within the thermal imagery domain, it is very, very important to understand the grey scale of the different type of detectors/cameras.
The thermal detector comprises of pixels in a horizontal and vertical array and it is positioned at equal distances from one another.
The detector material commonly used in all micro-bolometers within all thermal cameras is manufactured from Vanadium Oxide. Each and every pixel within a detector has a 14Bit code. Some manufacturers use 8 bits = 256 and another camera manufacturer uses all 14 bits which will give you 16384 scales of grey.
What does this mean?
To keep it simple, the one type camera can “see” 256 different types of grey per pixel within a temperature range and the other can “see” 16 384 different types of grey within a the same temperature range. This has a direct influence on the software detection capability or the so called video analytics. This is even more important when you have a scene where a human body and the background represent the same contrast as seen by your eyes. You cannot see the difference but the algorithm from the video analytics could, hence the reason why some analytics will detect better than others.
To explain it even in simpler terms, please have a look at the following images.
Now, as you can see above, the higher the number, the lessor amount of difference in contrast quality you will be able to see, next to one another. Needless to say, the difference between a greyscale of (2 to the power of 8 = 256) and (2 to the power of 14 = 16 384) is very big and the subtle difference between adjacent grey colours are of utmost importance. The selectivity can mean that you can either detect accurately or not. I have seen systems where the detection occurs and the eye cannot even see the individual being detected.
You can “see” the difference with Digital Detailed Enhancement Software. The heated lines is available in the first block, a little bit more detail is shown in the second block and even more detail in the third block.
There are so much more information available within a grey image, but dependent upon the detector type and make, it will be possible to have a much better image displayed and much better detection will be possible.
This is very important for video analytics as we need the selectivity and detail on the detector and due to the vast amount of data, for each pixel, it is much better to implement the video analytics on the camera DSP. This will ensure very detailed detection over long distances and the added value you get is that you can get much more information from a camera which in turn gives you a wider angle than the opposition. What does this confusing sentence mean?
Simply put, you can detect at 600m with a 50mm lens, just as well due to the advances on the pixel information. Other manufacturers need a 100mm camera lens to do the same detection at the same distance. As you know, a 50mm gives a much wider angle image on the horizontal plane than that of a tele focal 100mm lens of 4º field of view!
To conclude, if your detector and video analytics cannot “distinguish” between “clutter” and real activity, which will be driven by the selectivity and sensitivity of both mentioned items in combination, your detection will be bad or incorrect. Therefore, the more grey scales available, 2 to the power of 14 per bit = 16 384, the more images will be distinguished within the grey scale, the more and better detection will be the result, especially on perimeter fence detection systems.