With increased digitalization in the field of industrial image processing, the industry sometimes rashly writes off conventional technologies. In actuality, many users cling to their analog image processing systems. Those who take the needs of these users seriously have come to realize that the reason for this does not lie in analog transfer technology as such, but in the advantages of interlaced sensors, which are widely used in analog systems and previously were not available in cameras with a digital interface.

Analog vs Digital

Digital cameras, digital camcorders, MP3 players, and video on demand — when it comes to consumer electronics, digital technology pervades our households. Like the consumer market, the market for industrial image processing is characterized by increasing digitalization. Even so, digital interfaces are not catching on as easily and quickly in industry as they are in the consumer market. More than 50% of all systems still work with analog data transfer technology.

Figure 1. In the interlaced frame integration operation mode, there is no binning. In the first field, only the odd lines, and in the second field, only the even lines are read out. Therefore, while one line is read out, the next line can already be scanned.
Yet digital interfaces, such as the popular FireWire standard (IEEE 1394), offer considerable advantages for industrial image processing. They permit faster, more reliable image data transfer, without loss of quality, to a computer-based system that can process or archive the images directly, without a frame grabber. Furthermore, a digital interface permits multi-camera operation and comfortable parameterization of cameras.

Given all these advantages, why are many users of analog systems still so hesitant about switching to digital? Due to the life expectancy of machine vision systems, the migration to a new digital technology can obviously not be as swift as in the fast-moving consumer electronics sector. Furthermore, many users shy away from the costs involved in conversion, especially in the case of simpler analog systems where price is a major factor.

A more decisive reason is that almost half of analog systems work with interlaced cameras. For users of such systems, the switch to a digital interface previously translated into a twofold technological migration, because there simply were no cameras with interlaced sensors and a digital interface. Therefore, these users had to replace not just the interface, but the system technology as well. This is a complex and costly undertaking, because such a switch may have effects throughout the system. For example, new lenses have to be found, image processing has to be adjusted to the new image analysis, and the mechanism has to be adjusted as well (new mountings, new distances).

What is “Interlaced”?

Interlaced is an invention from the 1930s, when the development of television technology was taking its first steps. Many regard it as more of an art than a technology because it saves bandwidth for a given frame refresh frequency or increases the resolution. At the time, the task was to achieve a frame refresh rate of 50 Hz (in Europe) or 60 Hz (in America). No one spoke of 100-Hz televisions at the time.

The goal was to keep the picture from flickering in the eye of the viewer, with a maximum channel bandwidth of 6 MHz, which therefore can represent only 200,000 (blurry) pixels. The solution was to transmit each frame consecutively in two halves by having the scanning beam of the tube camera of the time and the corresponding electron beam of the tube TV skip every other line.

Applied to today’s technology with CCD image sensors and digital displays, this means that in the first image field, the odd (red) lines are read and drawn, while in the second field, the even (gray) lines are scanned and displayed. This explains the feat accomplished by the line skip: spatial resolution is replaced by temporal resolution. Interlacing them achieves the same resolution as a system with twice as many lines, at least for still scenes, and the image does not flicker due to the image fields’ high frame refresh rate of 50 Hz or 60 Hz, respectively.

Typically, these two fields are displayed sequentially on a monitor or, alternatively, de-interlaced in a PC and then combined to a singe full frame. However, the two fields are recorded consecutively; that is, at different times. This can result in artifacts, especially when the frame consists of moving objects. Certain tricks are necessary to make sure that the interlaced camera delivers good pictures in the intended application; for instance, by adjusting the timing and exposure settings, the lighting, and the image processing software.

Benefits of Interlaced

Figure 2. With color sensors with a CyYeGrMg filter matrix, up to 500% higher sensitivity could be achieved compared to progressive scan color sensors.
Why is this disadvantage not simply avoided by switching to a digital camera with progressive scan? There are two main reasons for this. The first reason is more a matter of economy. Extensive modifications in the construction for the timing and exposure control would be required to calibrate the system in a way that would make it work with frames recorded with a progressive scan sensor rather than an interlaced sensor. This involves a major effort and considerable expense, which for many users would outweigh the advantages.

The more important reason that keeps many users from migrating is based on the great light sensitivity of interlaced CCD sensors compared to progressive scan sensors. To understand this sensitivity better, we will now describe both modes of operation for interlaced sensors: interlaced field integration and interlaced frame integration.

The operation mode of interlaced field integration combines two lines vertically through binning, which already results in twice the sensitivity. Two pixels must share one memory bin or the memory bin has to store a lesser amount for each pixel and is therefore more compact. In the first and second fields, different lines are combined. This is also the difference of a progressive scan sensor, which is also capable of binning. After each field, the sensor memory bin is completely empty. In this mode of operation, both an external trigger and an electronic shutter are possible.

In the interlaced frame integration operation mode, there is no binning. In the first field, only the odd lines, and in the second field, only the even lines are read out (see Figure 1). Therefore, while one line is read out, the next line can already be scanned. As a consequence, each pixel can be scanned for a period of two fields. Put differently, both fields are integrated by overlapping. In the second instance, no external trigger or electronic shutter is possible; instead, a full-resolution frame can be achieved by using a flash during the overlapping.

Another advantage some conventional interlaced sensors offer is their greater sensitivity in the visible and near infrared (NIR) range. For example, the interlaced sensors in ExviewHAD technology for the visible and NIR range, which were developed by Sony, are among the most sensitive CCD sensors for consumer products. These sensors have already been very frequently installed in analog cameras for industrial image processing systems. Their sensitivity is further increased by putting one or even two microlenses on each sensor pixel so that more light reaches the light-sensitive zone of the pixel surface. With this technology, up to 300% higher sensitivity compared to progressive black-and-white sensors can be achieved.

Finally, interlaced sensors also offer advantages when it comes to color shots. Typically, a color-capable interlaced sensor receives a Bayer filter matrix with the colors CyYeGrMg (cyan, yellow, green, magenta) instead of the colors BGGR (blue, green, green, red) that are standard for sensors with progressive scans. Conventional CCD sensors are least sensitive at the two ends of the visible spectrum. The CyYeGrMg colors are closer to the center of this spectrum, which has the highest sensitivity. To be sure, this requires that the algorithms be adjusted in order to interpolate between the color pixels and to perform the white balance, but the resulting advantage – increased sensitivity – is striking. It is also interesting to note that with the complementary color sensors, binning is also possible with color. With color sensors with a CyYeGrMg filter matrix, up to 500% higher sensitivity could be achieved compared to progressive scan color sensors (Figure 2).

Interlaced Sensors with Digital Interfaces

Given these advantages, it is understandable that interlaced sensors for certain applications continue to have their fans. One example is applications with bad light conditions, such as endoscopy. Previously, migration from an analog to a digital interface was not an option for these users as only progressive scan sensors with a digital interface were available on the market. Now, there are products available that make the combination of interlaced sensor and digital FireWire interface available for the first time.

These products are equipped with the same interlaced sensors as most analog systems. This makes the migration to a digital interface possible without having to forego high sensitivity. Since the sensor technology remains the same, not only the interface, but also the rest of the system (lenses, lighting, and image processing software) require only minor modifications. Ideally, only the driver needs to be replaced. Therefore these new cameras protect earlier investments in interlaced-based systems, which they improve by using digital data transfer technology. Users no longer have to choose between the advantages of analog or digital technology; they get the best of both worlds.

This article was contributed by Allied Vision Technologies, Newburyport, MA. For more information, Click Here 


Imaging Technology Magazine

This article first appeared in the December, 2008 issue of Imaging Technology Magazine.

Read more articles from this issue here.

Read more articles from the archives here.