If you are like me, and do a little bit of photography in your spare time, I am sure that by now you would have some digital camera.
If you are a more demanding photographer, a digital SLR would be what you must have.
In the last few years we have completely replaced film photography with digital.

But, is digital really equal or better than film? Are any of these technologies as good as human eye?

Well, there are a few aspects of this question, and most of them apply to CCTV cameras as well, hence I am going to give you my opinion on it.

At the beginning of solid state imaging development in photography, the pixel count was a big limiting factor. Initially, CCD chips with only 100×100 pixels were developed. This improved considerably with the technology, and television was the first to see the benefit of it, as standard definition television (containing 720×576 pixels when PAL is digitised, or 720×480 when NTSC is digitised) was easily achieved.

Digital photography needs much more pixels.

We can all remember only a couple of years ago - 6 megapixel chips were so rare and expensive that only some pro cameras had. Today 10 and 12 megapixels chips in SLRs are very common, whilst medium size cameras have gone well over 30 megapixels.
Everybody knows that increasing the number of pixels increases the details of an image, that is probably the most important benefit, but if the lens is of inferior quality this may not necessarily be the case. Lens is very important especially with more megapixels. This is exactly the same issue with CCTV megapixel cameras. It is no good if you have the best megapixel camera and use some shonky piece of glass that pretends to be a lens. The image will be as good as your weakest part.

Another important consideration when going to more megapixels is the image file size.
More megapixels needs more processing time, more storage space and takes more time to handle it.
So it makes sense to find a good compromise. For example, if you know your largest photograph will not be any more than A3 format (300×420mm) and if you know that you can blow up a 10 megapixels image to this print size without visible pixelisation, than there is no reason to go much larger imaging chips than this. If you are a professional photographer, however, and you must have the best possible quality available, sure, sell your car and buy some decent 32 megapixel Hasselblad.

But despite all the megapixels and progress in technology, one major limitation of all imaging chips is the dynamic range. This is the capability of the chip to convert light information in one scene from the darkest to the brightest part without loss of details. If part of a scene is illuminated too low, chances are there are not enough photons coming from that area to be converted to a meaningful electronic signal. The detail in the dark will not be “seen” by the camera, especially if in the same scene you want to see the clouds in the bright sky for example. If a too bright area is projected by the lens (such as blue sky with clouds), chances are there are too many photons and the pixels become over-saturated with electrons, so that such an area very often appears as saturated white with no details. The illumination difference between a bright sky and a scene in the shadow can be as high as 100,000 lux to 0.1 lux. This is a ratio of 1,000,000 to :1. Expressed in dB, this is 120dB.

Human eye is capable of seeing see such a dynamic range of 120dB. This is a result of millions years of evolution in which the key feature is that human is a daylight being, thus adopting illumination ranges available from a bright sunny day to approximately full moon illumination. In addition, the eye cells have logarithmic response curve, making this distribution of 120dB, more efficient in low light levels.

Unfortunately, CCD and CMOS chips do have big limitations here. An average maximum ratio between the brightest scene area and darkest, in one shot, is only around 3,000:1. This represents a dynamic range of 60dB. Way too small for a typical daylight scene. So, correct exposure in photography is very important, and often it is dictated by the intention of the photographers. If he/she wants to show details in the shadow, the composition of the image would be such that bright sky will not be in the view, and appropriate exposure and iris opening (F-stop) is used for the shadow details to be exposed correctly. It is impossible to take a shot with digital camera and have good details in very bright and the dark areas simultaneously. This is the reason why the so-called High Dynamic Range (HDR) photography was invented. Basically what this does is take a couple of shots of the same object, but with various exposures (so-called “bracketing”): normal, overexposed (which will show details in the darker areas, and underexposed (which will show details in the very bright areas). There is some specialised software which combines these two or three shots in one, with some clever blending and you get a shot that has details in all ranges. Even latest PhotoShop has HDR function for some basic high dynamic output. If you didn’t know about this search under HDR photography and you can see some amazing sample photos made this way.
So, in photography, HDR is possible but only after processing images after they have been taken with various exposures and/or F-stops.

In CCTV we have a bit of a problem because the exposure is fixed, 1/50s (for PAL) or 1/60s (for NTSC). F-stop (iris) is usually controlled automatically, but with fixed exposure so that the dynamic range becomes even smaller than in photography. Yet, there are some inventive ways and they are in essence like using “bracketing” and superimposing (“double-exposure”) two differently exposed TV fields. The first to start this trend was Panasonic many years ago with their wide dynamic range cameras. The limiting factor is that the exposure can not be longer than the TV field exposure, but only shorter. So this is typically done with 1/1,000 or shorter exposure superimposed electronically with the 1/50 second exposure. This is the so-called wide dynamic range, being processed 50 times each second for PAL, or 60 times for NTSC.
Many brands on the market are using this “trick,” of merging two different exposure into a high dynamic range.

Wouldn’t it be nice however to have an imaging chip that will genuinely have wider dynamic range?

Of course it would be nice. But there are a number of limiting factors, which with the current technology are not possible to change. First and most important limitation is the physical pixel size. The smaller the pixels are the smaller the dynamic range. The demands of the market to have smaller cameras and smaller lenses forces the manufacturers to make imaging chips which border with the impossible. For example, today we have 1/6” chips on some cameras, yet preserving the standard definition of 400 TVL or more. Clearly this leads to pixels being so small that the accumulating holes are very small.

The second limitation is the working temperature of cameras. At any temperature above the absolute zero (-273˚C), by definition, there is always such activity in the semiconductor layer that electrons generated from heat is significant, thus represents unwanted signal – thermal noise. Thermal noise, like any other noise, reduces the dynamic range. This is one of the most important reasons why in astronomy for example, cameras and their imaging chips are cooled down to -40˚C or lower with special cooling devices, thus reducing the thermal noise to minimum.
In CCTV, clearly, we can not afford such measures and cooling for chips is not even considered. It would be beneficial however to know that each 3˚C add about 50% of the RMS noise. Thus, making an effort to reduce camera direct exposure to sun may help a lot.

Having said all of the above, there is a new and smart technology that actual works differently and it does offer a quantum leap in achieving dynamic range of almost 120dB without the increase of the pixel size. This technology is patented by Pixim, and I will talk more about it in one of my next writing…

• BROWSE / IN TIMELINE
• « Video or Image compression ?
• » Psychophysiology of viewing CCTV

• BROWSE / IN CMOS Cameras CCD