We have talked a lot about improvements in CMOS image sensors in low light imaging capabilities which are particularly helpful in outdoor applications such as long-range observation, surveillance, and global security. The image sensor parameters relevant to low light are QE (Quantum efficiency), read noise, dark current, uniformity, and MTF (in NIR).
For all of these parameters, CMOS image sensors can now have better performance than CCD. Many security applications are even able to replace the systems daylight and lowlight cameras with 1 CMOS color uncooled camera.
As far as blooming and smear is concerned, CMOS has always had the advantage with the pixel architecture differences (no shift registers). Smear is well known as a vertical stripe in the image from very bright spots, but what is not widely known is that smear can also reduce the dynamic range and limit the ability to see details in the image.
Dynamic range, which is the ratio between the maximum output signal level and the noise floor (noise floor which is the RMS (root mean square) noise level in a black image) is very important for outdoor scenes. Not only is a low noise floor and uniform dark image crucial for good low light sensitivity, a low smear level is very important for imaging in bright daylight conditions!
We did a test set-up to investigate the effects of smear comparing CCD and CMOS and show how smear deteriorates your system performance and is more than just a cosmetic artifact.
Smear is most likely to affect your image quality in two conditions:
- Short sensor integration time (e.g. fixed aperture lens systems in bright daylight conditions)
- Strong overexposure of parts of the image (reflection of sun, laser beam, strobe light,…)
The Adimec test set-up to investigate the effects of smear with CCD and CMOS cameras
We gathered some images of a USAF target with the conditions mentioned above with a HDTV CCD and a CMOS camera.
CCD Camera CMOS Camera
At first glance, there is not a big difference – but let’s take a closer look. With the CCD camera, the smear is visible above the dark rectangle (darker band), but also in the dark area (black shift towards grey)
Though the light level is very high, this is not yet worst case:
- When the vertical size of the dark object is reduced, the smear will increase
- An image scene with a reflection of the sun may introduce higher smear levels!
In order to see what smear does to your image, we’ve stretched the video signal (offset and gain for the CCD, only gain for the CMOS sensor).
Detailed CCD Camera Image Detailed CMOS Camera Image
In the image from the CCD, the background is not very uniform (smear depends on image content per column, black level is dependent of image data on horizontal lines). But even more important: the word THORLABS is difficult to read due to noise. Fine patterns are difficult to resolve (bottom right part of the image).
It is important to understand that the noise in the CCD image is not caused by the read noise of the image sensor alone. A large part of the noise that you see is shot noise from the smear in the image! (Smear is causing an image dependent DC offset. Though you can compensate for this offset in your image processing, the shot noise in this DC offset remains…)
To summarize – smear causes:
- Contrast reduction of the image
- Questionable patterns depending on image contents
- Higher noise level in the dark areas of the image
- The three effects above combined: degradation of DRI (detection, recognition, identification)!
- For color: a shift of white balance, depending on the smear level
This is why you will have a harder time to see (low contrast of fine) details with a CCD camera in such situations. Adimec cameras in the TMX series that are using the latest CMOS image sensor technology outperform the ones using CCD sensors!
Note that this example is using a high contrast target. In real, low contrast situations the problems will be even worse!! Any comments or additional ideas?
Related Blog Posts: