SECTION II

Chapter 5
CCD Image Capture Devices

What This Chapter is About

Charge-coupled devices (CCDs) are solid-state semiconductor circuits that detect impacting photons ranging in wavelength from the near infrared to X-rays. They were conceived at AT&T Bell Labs in 1970. Depending on their design, CCDs can be extremely sensitive. Some commercially available CCDs reliably detect fewer than ten photons, allowing phenomenal images of extremely faint objects. Since CCDs detect nearly every arriving photon and produce linear output signals (directly proportional to the number of detected photons), CCDs are considered nearly perfect photon detectors.

As the heart of digital imaging devices such as desktop scanners, camcorders, and digital cameras, CCD applications range from astronomy to zoology, A to Z. CCD devices are rapidly becoming ubiquitous. The principles behind CCD design and operation are both interesting and instructive in understanding their strengths and limitations.

Monochrome capture devices require one CCD while color capture devices require color filtering. In one color capture approach, a capture device uses one CCD and requires multiple exposures, one through each filter. Another more expensive approach requires one exposure and uses three CCDs that are positioned behind one of three colored filters. An alternative approach requires one exposure and uses one CCD that is positioned behind a filter with different colored areas. This trades spatial resolution for color information.

In any event, color filtering introduces many previously discussed subtractive color problems. Finally, because color capture reduces to three individual monochromatic image captures, the remaining text only considers monochromatic image capture.

Imaging Fundamentals

To capture a digital image first requires partitioning a desired view into numerous, distinct sample areas. Each sample area is usually rectangular or square. Sample areas collectively encompass the view and must be small enough to capture desired detail. In general any significant detail transition must span at least two sample areas. It follows that highly detailed views require more sample areas for faithful capture operations.

After exposing the collective sample areas to a scene, each sample area result is encoded by assigning an average quantized value to represent the total sample area. Encoding values usually range from 0 (representing black) up to a maximum value (representing white) that is usually convenient for computer storage (1 for bilevel images or 3, 7, 15, 31, 63, 127, 255, 511, 1023, 2047, or 4095, etc. for monochromatic images). These collective values allow subsequent view reconstruction after they are saved in an image file.

CCD Imaging Basics

CCD construction is well suited to digital imaging. In a CCD imaging device, a lens focuses views onto a CCD which contains an M x N photosite (sampling) array. Individual photosites are often called CCD pixels or just pixels. An M x N photosite array may be collectively regarded as M linear registers, each containing N photosites. Channel stop regions isolate each linear register from adjacent linear registers and polysilicon gate structures isolate photosites within a linear register from adjacent photosites. This allows each photosite to sample independently of any other photosite. In Figure 5-1, the physically adjacent M linear registers are vertically aligned.

During image sampling, the photoelectric effect allows CCD photosites to detect arriving photons. An arriving photon strikes a photosite, creating one or more electron-hole pairs within the silicon structure. This creates an electrical charge that accumulates within the photosite with each impacting photon. The channel stop regions normally prevent accumulating charges from leaking into adjacent photosites. The cumulative charge indicates the number of incident photons over the integration time, the sample period.

Figure 5-1  Photosite (sampling) array

After the integration time expires, the M x N accumulated charges must be transferred from the CCD to the imaging circuitry. This requires an additional independent linear register, the serial register, placed next to the M x N photosite array as in Figure 5-1.

To use the serial register, the M linear arrays transfer the weak photosite charges closest to the serial array to corresponding elements in the serial array. The serial array then sequentially feeds each transferred charge into a charge detection output amplifier which produces an amplified voltage. Other logic uses this voltage to produce an appropriate digital pixel sample value for presentation to other components.

During the serial amplification process, the CCD linear arrays carefully ripple their charges to the next pixel position in preparation for the next serial register transfer operation. The resulting digital values represent the view focused on the CCD. Figure 5-2 illustrates the process.

Figure 5-2  Serial amplification process