Here is how digital camera designs capture images that include motion.
The Part 1 of this series about digital camera design has shown how to do the basic calculations that interrelate pixel size, focal length, focal ratio, and spot size, and demonstrated how these parameters can affect exposure time and image size and, ultimately, image quality. Part 2 will build on that knowledge to examine how cameras capture images that include motion in a scene, looking at different types of mechanical and electronic shutters and the differences in how they operate.
Exposure time and frame rate
We have all seen images with moving objects in the scene that look blurry (Figure 1). Because a camera needs a finite time to expose an image, any motion occurring during that time tends to smear the image features over several pixels (Figure 2).
Figure 1 A photo taken of a moving object looks blurry when the image features are smeared over several pixels. Source: Etron
So, if there is motion, then one way to reduce the smearing is to make the exposure time short relative to the velocity of the image moving on the focal plane. Unfortunately, there are limits. The exposure cannot be arbitrarily shortened as there may not be enough exposure time to capture an acceptable image considering the focal ratio, camera sensitivity, and scene lighting intensity.
To engineer a solution with acceptable results, we need to quantify the problem. Let’s start with a list of questions:
Let’s assume we can tolerate the spot moving over two pixels during the exposure. Let’s also assume we use the same imaging system from Part 1: 2-micron pixels, a focal length of 10 mm, and the target is at the same 100-meter range. Let’s further assume the velocity of the object is 10 meters per second perpendicular to the camera. So, what is the maximum exposure time we can tolerate and still keep the motion smearing no more than two pixels?
In Part 1, we learned how to calculate the IFOV as a function of pixel size and focal length. For our example of a 2-micron pixel and a 10 mm focal length, the IFOV is 0.2 milliradians per pixel, and that means at 100 meters, a pixel covers 20 mm.
Since, in this example, we specified the target was traveling at 10 meters per second at 100 meters away, in pixel terms it is traveling across the sensor at 500 pixels per second—10 meters per second/20 mm per pixel. Since we imposed a limit of no more than 2 pixels smear, that limits the exposure time to 1/250 second or 4 milliseconds.
If we were creating a video of this scene, the 4 milliseconds for the maximum exposure time would correspond to 250 frames per second (FPS) as the minimum frame rate required to limit the smear within a single frame to no more than 2 pixels when the object is traveling 10 meters per second at a 100-meter range.
Let’s examine how to start and stop the exposures. When exposing a focal plane array to a scene with motion, the way the exposure is started and stopped for each pixel becomes important. The design of the pixel and corresponding sensor readout architecture can introduce artifacts such as smearing or distortion during readout. Moreover, mechanical shutters can cause both distortion and shading artifacts under certain exposure conditions. Pulsed light sources can avoid the problems mentioned but are not practical for all applications.
Mechanical shutters include iris type shutters (Figure 3), shutter in wheel configurations (Figure 4), and curtain schemes with a sliding shutter (Figure 5). Each scheme will have different phases in its open-close cycle: 1) close to open switching time, 2) open time, 3) open to close switching time, and 4) close time.
When the switching times become more than about 10% of the time the shutter is left open, then there is the potential for “shutter shading” when using iris-type shutters. The center is simply open longer than are the edges leading to unequal exposure time. The larger the sensor relative to the iris clear diameter, the more of an issue it becomes.
Wheel and curtain type shutters can equalize the exposure time for each pixel since they sweep uniformly over the sensor at both beginning and end of exposure. However, in all the mechanical shutters, the potential for focal plane distortion arises. The start and end time of the exposure of each pixel can vary, even though the shutter design of each one of them is potentially exposed for the same time.
Because the start and end times for each pixel are different in the case of the wheel or curtain shutters, focal plane distortions can occur when motion is taking place in the image frame. This is true for film and digital cameras and dates back to the earliest days of photography. The image in Figure 6 is an early example of focal plane distortion that would be observed in the curtain shutter of Figure 5.
Note how the bottom of the image was “shuttered” earlier than the top part, leading to a warping of the rapidly-moving vehicle. The standing observers and vertical poles have no such distortion since they are not moving; only the racer is moving and only the racer is distorted.
When the brightness of the scene changes rapidly during the shuttering, other artifacts can be observed when using mechanical shutters. A good example is a lightning flash that occurred as a curtain shutter was in transition (Figure 7).
Iris shutters, if used to take very short exposures, will leave artifacts in the image. Figure 8 shows an example of an image taken using a camera that has both an electronic and mechanical iris shutter. The image on the left shows the mechanical shutter used to take a 10-millisecond exposure. The right side shows the same exposure duration, but with the mechanical shutter open and using the electronic shutter to time the exposure.
Note how the mechanical shutter left a non-uniform pattern in the image versus the uniform image when the electronic shutter was used. No lens was used on the camera; it was aimed at a dim ceiling. The transition time of the shutter was interfering with the image, leaving an illumination artifact in the image.
Continue reading on EDN US: Electronic shuttering
Richard Crisp is VP of New Product Development for Etron Technology America.