CAPTURING IMAGES WITH DIGITAL STILL CAMERAS
WE FACE MULTIPLE TECHNICAL TASKS WHEN DESIGNING AND DEVELOPING
DIGITAL CAMERAS FOR CONSUMER USE. AS WOULD BE EXPECTED, IMAGE QUALITY IS A TOP CONCERN.
Shoji Kawamura Olympus Optical Co. Ltd.
Digital still cameras fall into two categories. One uses a black box and a camera lens with a CCD (charge-coupled device) back unit that can acquire more than 3 million pixels. Designed for professional and business use, this camera normally costs more than $5,000. The second type of digital camera is designed for consumers and employs a CCD capable of acquiring 350,000 to over 1 million pixels (Mpixels). Because it is easy to handle and carries a reasonable price for the picture quality, its market has grown dramatically in last three years. Here, I discuss the image-capturing system for a consumer-use digital still camera after briey explaining its history.

Consumer digital cameras

Until 1981, consumers had to use an analog still camera to convert image light signals into electric signals. In 1981 Sony introduced its Mavica system, which recorded images on magnetic diskette, and the history of digital still imaging systems began. At the 1988 Photokina (one of the biggest photo exhibitions in the world, held every two years in Cologne, Germany), Fuji Photo Film announced the rst digital still camera. The Fujix DS-1P used a memory card for recording image data. At the following Photokina 90, Fuji Photo
Film and Olympus exhibited the prototype of todays digital still cameras. These cameras used an ASIC (application-specic integrated circuit) to cope with the compression algorithm that was standardized by the JPEG (Joint Photographic Experts Group) draft standard. Their configurations were almost similar to current digital still cameras. Apple introduced the QuickTake 100 as an image-inputting device for computers with very aggressive pricing in 1994, and in 1995 Casios QV-10 experienced big sales with its LCD monitor and attractive price. These models marked the beginning of a new era in consumer-use digital still camera. At Photokina 96, Olympus introduced its XGA (extended graphics array) C-800L that could process 810,000-pixel images. Until then, only 350,000- to 450,000-pixel VGA (video graphics array) cameras that produced an image quality far lower that of silver-halide (film) cameras were available. The XGAs improved image quality, approaching that of the silver-halide images. In 1997, Olympus introduced C-1400L; the rst digital still camera that included an over 1-Mpixel CCD (total 1.4 million, effective 1.3 million) for the consumer market. After this models introduction, many Mpixel cameras arrived for the consumer market. Before the introduction of the C-1400L,

0272-1732/98/$10.IEEE

VGAs were the mainstream choice. They used a video CCD capable of acquiring 400,000 pixels but produced a lower image quality than silver-halide cameras. To produce a similar image quality, VGAs added an over 1Mpixel CCD, 1,280 1,024-pixel SXGA (super XGA) image size, as well as a highresolution lens to cope with the CCD. SXGAs became indispensable. C-1400Ls also used a specialized color algorithm and received a good reputation for capturing silver-halide image quality levels.

Digital camera

CCD (a)

Film (b)

Conguration
The physical difference between digital still cameras and silver-halide cameras is their recording process: on an image sensor or on silver-halide lm. (See Figure 1.) There is a big difference in how the two cameras make images visible. Silver-halide cameras (Figure 1 shows a single-lens reflex camera) lead the light ray through the lens toward the film. The film base reacts to the light, and the image data is exposed on lm. This processing takes place in the camera body. After these steps, processing involves taking the lm from the camera, developing it with chemical materials, and lighting the lm with three kinds of color lights. A digital still camera leads the light ray through the lens toward the image sensor, focuses the image on the sensor, and converts the light value into an electric signal. The electric circuit (an analog-digital converter) converts the signal into a digital value and adds color signal information to the digital value. To reduce the data size, the camera normally compresses and records the data into built-in semiconductor memory or into removable memory. Since many of these cameras have LCD monitors, the user can check the image immediately on the monitor. In addition, users can play back the image on a TV monitor via the cameras video output terminal. To handle the image data, the hardware and software in the camera share the burden to lessen the number of electrical parts; this shortens the processing time and keeps the exibility for adjustment parameters. For this purpose, digital still cameras require a high-performance RISC (reduced instruction set computer), a CISC (complex instruction set computer), and hard-logic (an ASIC). (See Figure 2.)

Figure 1. Image recording in digital (a) and lm (b) cameras.

DRAM I/O control

Video out Video LCD I/F Coprocessor Video monitor

Memory manage

LCD monitor
RISC JPEG engine CCD timing I/F

Cache SRAM

PCMCIA I/F IRQ Clock

Flash media

Figure 2. Digital camera conguration. IF: interface; IRQ: interrupt request.
A digital still camera transfers the image onto an image sensor that is only 1/2 inch or 1/3 inch in diagonal size, while a silver-halide camera uses a 36 24-mm image area. As a result, digital still cameras required many technical breakthroughs to reach the high-quality images with such small sensors.

High image quality

In general, the CCD, which acquires larger numbers of pixels, produces the better quality images. Because of that, designers sometimes indicate image quality by talking about the number of pixels. Note however that the image quality varies by the type of CCD, color lter, signal-processing method, and optical system used. Therefore, simply stating the

NOVEMBERDECEMBER 1998

IMAGE CAPTURING
number of pixels does not accurately indicate image quality. The units in digital still cameras that inuence image quality are
lens, LPF (low-pass lter), CCD, processing circuit and algorithm, analog compression/decompression, and monitor and printer.
In addition to these factors, the focusing system and exposure control are also important for image quality. The lens designer must consider MTF (modulation transfer function; contrast character) and color characteristics. The lens is the most important unit in producing a quality captured CCD image. (more pixels) The LPF extracts noise from constant light. The adjustment of cutoff frequenCCD CCD cy is important in that a lower CCD size: as is CCD size: larger Cell size: smaller Cell size: as is cutoff frequency setting lowers image quality, and a higher cutoff frequency setting CCD CCD Task produces a moir, or rippling, Better S/N ratio Lower cost of an images color. The CCD is the eye of a digLens Lens ital still camera. The MTF Larger image Task Higher resolution, characteristics of the CCD and larger f number, circle keeping less lens resolution the characteristics of the color aberration and aberration lter (which is placed in front of the CCD) greatly inuence image quality. Figure 3. Raising resolution. The processing circuit hanXGA (0.80 Mpixels) 1/3 in. 1in. 2/3 in. 1/2 in. 1/3 in. 1/4 in. 8 in. 1998 Year (a) (b) SXGA (1.3 Mpixels) 4.7 m 1/2.7 in., 4.15 m 3.7 m 3.2 m 2.8 m

dles the electric signal according to the CCD characteristics and the combination of CCD and lter, thus obtaining the best image signal. Because image data sizes are increasing, cameras need extensive memory to save raw data. The analog compression/decompression circuit uses a compression algorithm to reduce image data, so less memory is required. The compression ratio is adjustable. A higher compression ratio produces a lower image quality but requires the least amount of memory; on the other hand, a lower compression ratio produces higher quality images and requires the most memory. Many digital still cameras adhere to the JPEG compression format, allowing images to be compatible in many different applications. The speed of the processing circuit and compression/decompression circuit affects the total processing speed of the camera, which is important factor for the cameras specications. The monitor or printer is the tool that lets users see the final image. If the monitor or printer does not cope with high resolution, it cannot reproduce a high-quality image even though the image has enough image data. There are many factors we can use to evaluate image, color, distortion, shading, ghosting, contrast, resolution, and so on. The inuences of these factors are connected with the above-mentioned units.

Technical tasks

Designers have to complete many technical tasks in their development of digital still cameras. One of the important tasks is combining high-pixel CCDs and high-resolution lenses (Figure 3).

Using larger CCD chips

The difference between a 350,000-pixel CCD and a Mpixel CCD is easily recognized, as seen in Figure 3. (The right side of the gure shows the larger chip.) If the total pixel number is the same, the larger CCD makes one pixel bigger and achieves higher sensitivity than a smaller CCD. In addition, the requirement for lens resolution is less severe than for a smaller CCD. On the other hand, a bigger CCD lowers production efciency, thus raising the cost. A larger image circle requires a larger diameter lens, while keeping the level of resolution and
Figure 4. CCD size (a) and pixel size (b) versus cost.

IEEE MICRO

aberration stable. It makes a lens expensive, and it lessens the cameras portability.
1.0 0.9 0.8 0.7 Modulation 0.6 0.5 0.4 0.3 0.2 0.50 f22 f11 f5.6

Defocusing -0.02500

Using smaller CCD cells
Meanwhile, a smaller, less costly CCD (seen on the left side of Figure 3) and lens system reduce the cameras cost and size. Because of the pixels smaller size, it is necessary to improve the S/N (signal-to-noise) ratio to impose a higher requirement on the lens so it collects enough light (higher aperture number) and to focus the image on small pixels. Finding a good combination of CCD size, high resolution, and bright lens is one of the designers important tasks.

150 Spatial frequency (cycles/mm)
Relation between CCD size and cost
The CCD is one of the expensive electrical parts in digital still cameras; there is a close relation between the CCD size and the camera cost. A popular production process cuts out a CCD chip from an 8-in. wafer. A smaller CCD reduces the cost of one chip. For example, compare 1/3-in. and 2/3-in. CCDs. A 1/3-in. chip is one quarter the size of a 2/3in. chip, but costs one-eighth less. On the other hand, the production technology for semiconductors is an important key factor in acquiring small pixels on the chip. One-third-inch XGA (0.8-Mpixel) CCDs are produced with 4.7-m pitch. If the pixel number in SXGAs increases to 1.3 Mpixels in 1/3in. CCDs, it is necessary to establish a 3.7-m pitch production technology. Unfortunately, this technology is not yet at a commercial level; a 1.3-Mpixel CCD is in mass production in 1/2.7-in. by 4.15-m pitch. A 1998 announcement reported a 2-Mpixel CCD produced in 1/2 in. by 3.9-m pitch technology. Obviously, the improvement of semiconductor technology (see Figure 4) is accelerating the cost reduction of CCDs. CMOS technology is another factor to consider when developing a digital camera. Since CMOS production costs much less than a CCD, a cameras cost is therefore less. Battery consumption is also much lower. The technical breakthrough came in the development of the noise lter: The N/S ratio of a CMOS sensor is very low, and the lter deletes noise from signals. However, the current lter level is not good enough for a digital camera, though I expect it will soon improve.
Figure 5. Aperture size versus diffraction.

The lens

Lens designers also must cope with several tasks when designing high-pixel CCDs. When the pixel size is small, sensitivity is less and the diffraction by the aperture grows. The small pixel size requires a lens setting with a bigger f stop (aperture size) to avoid diffraction. Figure 5 shows the relation between the lens f number and diffraction. The modulation equals the lens resolution power against spatial frequency. With low modulation, the lens cannot produce an image because of the diffraction. For example, if we take a picture with the spatial frequency set at 120, the lens cannot produce an image when set at f22. Figure 6 (next page) shows the difference between setting the same focal-length zoom lenses to f1.8 (Figure 6a) and f1.2 (Figure 6b). A lens set at f1.2 must be 10% longer in length and 15% bigger in diameter, and include one more element. It is important for a high-pixel CCD to have a brighter lens in a smaller size.

The CCD

There are technical tasks for designing a digital still camera that come from the CCDs characteristics. We have two types of CCDs for these cameras, one is an interlaced-scan CCD mainly used for video cameras. The other is a progressive-scan CCD developed exclusively for the digital still camera (Figure 7).
Developed for outputting moving images to TV monitors, the interlaced CCD scans odd lines rst and then even lines. It combines

y z 2.00 X

f1.2 (b)

z 2.00 X

Figure 6. Comparing an 8X zoom lens (from 6.5mm through 52mm). An f1.8 setting can make the lens 10% shorter, 15% smaller than f1.2, and requires nine elements (a). An f1.2 setting (b) requires 10 elements.

Sensor Signal charge

Vertical register (a) (b)
Figure 7 Progressive- (a) versus interlaced-scan (b) CCD using electrical and. mechanical shutters.
the data of the two fields into frame image data. In this case, as the whole frame is made from data derived from the two fields and there is a time lag, the moving object cannot be stopped simply by scanning. That is why the interlaced CCD in many digital still cameras requires mechanical shutters. A progressive CCD scans the entire frame image data at once. An electrical shutter can capture the image, if the pixel number is not big and the frame rate is fast enough. However, if the frame rate is relatively slow with a Mpixel CCD, a smear phenomenon can easily occur. (As a strong light is charged on pixels, even in a short period during transformation, a white line comes in from a vertical direction.) To avoid smearing, designers need to add a mechanical shutter to progressive-scan CCDs. For conguration, the interlaced-scan CCD requires a narrow vertical register, because the electrical charge is not big. It makes pixel cells larger and more sensitive than progressivescan CCDs with same number of pixels. The combination of shutter speed and f setting of the lens denes the exposure control program; both speed and f setting limit this program. The CCD has the same sensitivity as silverhalide lm, as dened by ISO (the International Standards Organization) standards. The latitudewhich shows the allowance level at which a CCD can detect darkness and brightness levels for this ISO standardis called dynamic range. The dynamic range of a CCD is generally narrower than film. The small dynamic range of a CCD requires more strict exposure control than a silver-halide camera.
Propotional dimension a, b, d, D, D a > xxx (for production) d > xxx (for production) Constant dimension = 0~0.015 mm n : lens index R Tolerance of all dimensions
Worse : angle of lens axis Evaluation of small d, small r
Figure 8. Using a smaller lens size (a) and its inuence on production (b).

Mass production

Today, designers also face the technical tasks for mass production of image-capturing systems. As mentioned earlier, a small and bright lens is necessary to produce correct images on high-pixel CCDs. However, even though designers acknowledge the characteristics of the lens and CCD, the mass production steps will still vary the nal image quality. Several parameters affect lens quality. Designs automatically determine parameters a, b, d, D, and D. Figure 8a is a cut drawing of a single sample lens showing these parameters. Other parameters must be controlled in mass production; otherwise lens capability will not be delivered as designed. The most important parameters are the angle of lens axis and the shift of axis (Figure 8b). They must be strictly controlled in mass production. In conclusion, designers must consider many important units when designing and developing digital still cameras (see Figure 9). We considered all these important factors when we designed and developed our Mpixel camera, which can acquire 1.4-Mpixel images on a 2/3in., progressive-scan CCD (as shown in Figure 10) using a 3X, f2.8-to-f3.9 zoom lens. This camera has achieved a high reputation in consumer markets.

Lens Image circle size Area size 3x zoom f2.8-3.9 Cell size

CCD S/N ratio

f number

Color filter

Aberration

Progressive/ interlaced

Resolution
Productivity Size: length, diameter

Diffraction

Productivity
Figure 9. Important units to consider for digital still camera design and development.
he introduction of a digital still camera that combines a multiMpixel CCD and a high-resolution lens, and is sold at a reasonable price, has changed the consumer digital imaging market. The image quality is reaching an almost similar level as the silver-halide camera, allowing the digital still camera to sell well on the consumer market. At the same time, since the image size of these cameras is increasing, we expect to see the market for smaller cameras also increase because they handle data easily. We will see two categories of digital still cameras, one to produce high-quality images and another for use as an image-inputting device for computer peripherals. Both lines will be improved and expand their markets in the future; they will also add fascinating digital features. MICRO Shoji Kawamura is manager of the Digital Imaging Business Development Department at Olympus Optical Co. in Tokyo. Earlier, he worked in the Silver Halide Camera Research
Figure 10. The Olympus 1.41-Mpixel, 2/3in., progressive-scan CCD.
and Development Department and served in Germany as a technical liaison of the R&D Department. Kawamura received a bachelors degree in electrical engineering from the Waseda University in Tokyo.
Direct questions concerning this article to Shoji Kawamura, Olympus Optical Co., Ltd., DI Business Development Department, SanEi Building, 22-2 Nishi-Shinjuku 1-Chome, Shinjuku-ku, Tokyo 163-8610, Japan; s_kawamura@ot.olympus.co.jp.

PixController Universal RS-232-U User Switch Settings
Copyright , PixController, Inc. http://www.pixcontroller.com, all rights reserved. The User Control Switch (SW1) will let you customize how the RS-232 Camera Control Board will trigger the Digital Camera. Here you can adjust the time delay between pictures, operating only at day, night, or 24 hours, setting up a walk-test mode for testing PIR range/area, set the type of digital camera to control, and turning the control board LEDs on or off. The advantage of this type of unit is that you simply connect your camera to the board with a Serial Cable (supplied with your camera). Below is a list of all cameras supported by this board. Note: When turning power on to your Universal board both the red and green LED will light up. They will both stay on for 30 seconds. This time will allow the PIR circuit to warm up. After this time expires the green LED will turn off and the red LED will blink 5 times letting you know that the board is entering a 1 minute automatic walk-test phase. At this point you can move around the camera setup and check out the PIR area. Both the green and red LEDs will light when motion is detected. After the 1 minute automatic walk-test phase expires the red LED will blink 5 times letting you know the camera system will now become active.
Types of Digital Cameras that can be used
Listed below are the types of digital cameras the Universal RS-232 board. Devices are connected to the Control board VIA the Cameras serial cable (plug-n-play), and do not require any camera modifications. This is a great way for a beginner in the trail camera building market to start out. The cameras on this list have not all been tested, but most of the Olympus cameras should work. The Universal RS-232 requires that the camera meets the RS-232 spec of +/- 12V generation. If the camera does not meet the RS-232 spec there is a good chance the camera will not function properly with the Universal RS-232 board. The Olympus D-340R and D-360L have been tested and work properly. We recommend these cameras for trail camera use with the Universal RS-232 board. Even though these cameras are no longer made they can easily be found and purchased on eBay. Do a search on the description for Olympus D-360L to locate a camera. If you find that a camera does work please email us as support@pixcontroller.com and we will make note of it for other users.
RS-232 Sierra Mode Digital Cameras
Bold Red cameras that should work Italic Gray untested cameras Agfa ePhoto 1280 Agfa ePhoto 1680 Agfa ePhoto 307 Agfa ePhoto 780 Agfa ePhoto 780C Apple QuickTake 150 Apple QuickTake 200 Chinon ES-1000 Epson PhotoPC 3000z Epson PhotoPC 500 Epson PhotoPC 550 Epson PhotoPC 600 Epson PhotoPC 700 Epson PhotoPC 800 Nikon CoolPix 100 Nikon CoolPix 300 Nikon CoolPix 700 Nikon CoolPix 800 Nikon CoolPix 880 Nikon CoolPix 900 Nikon CoolPix 900S Nikon CoolPix 910 Nikon CoolPix 950 Nikon CoolPix 950S Nikon CoolPix 990 Olympus C-1000L Olympus C-1400L Olympus C-1400XL Olympus C-2000Z Olympus C-2020Z Olympus C-2040Z Olympus C-2100UZ Olympus C-2500L Olympus C-2500Z Olympus C-3000Z Olympus C-3030Z Olympus C-400 Olympus C-400L Olympus C-410 Olympus C-410L Olympus C-420 Olympus C-420L Olympus C-800 Olympus C-800L Olympus C-820 Olympus C-820L Olympus C-830L Olympus C-840L Olympus C-900L Zoom Olympus C-900 Zoom Olympus D-100Z Olympus D-200L Olympus D-220L Olympus D-300L Olympus D-320L Olympus D-330R Olympus D-340L Olympus D-340R Olympus D-360L Olympus D-400L Zoom Olympus D-450Z Olympus D-460Z Olympus D-500L Olympus D-600L Olympus D-600XL Olympus D-620L Panasonic Coolshot NVDCF5E Sanyo DSC-X300 Sanyo DSC-X350 Sanyo VPC-G200 Sanyo VPC-G200EX Sanyo VPC-G210 Sanyo VPC-G250 Sierra Imaging SD640
Modes of Switch Operation

Default Setting

All switches UP except switch 9 which is DOWN. 10 Second Delay between pictures, 24 Hour Recording, Sierra-Type Digital Camera Mode, Control LED On, PIR LED On. Red indicated the switch position for all graphics below.
Delays Between Pictures Setting
Switches 1, 2, and 3 control the delays between pictures.

10 Seconds Delay

20 Seconds Delay

45 Seconds Delay

1 Minute Delay

2 Minute Delay

5 Minute Delay

10 Minute Delay

20 Minute Delay
Day/Night Operation Settings

Switches 4 and 5 control Daylight, Night Time, and 24 Hour recording or pictures.

24 Hour Operation

Night Only Operation

Day Only Operation

PIR Walk-Test Mode
On boot up of the Universal board this setting will put the unit into a PIR Walk-Test mode. Here you can check out the PIR detection area without having the unit take photos. When booting the Universal board into this mode the RED Control LED and Green PIR LED will stay on for about 30 seconds. This is when the PIR is warming up. After this period of time has expired you are free to walk and test the PIR area. Note: To put the PixController back into Photo Taking Mode change the switch settings of switch 4 and 5 to one of the three options above under the Day/Night Operation Setting, and power the PixController unit Off and On from the external power switch.
Digital Still or Double Photo/Movie Mode
Switch 6 sets the type of Digital Camera that is connected to the Camera Control Boards RS-232 serial port.

Still Mode

Double Photo or Movie Mode

Movie Length

Switch 7 controls the delay between photos in double photo mode, or the length of the movie taken if the camera is setup in movie mode.

10 seconds

15 seconds
Control LED On/Off Setting
Switch 8 sets if the Control LED (Red LED) is to be used or not. Note, the control LED will always be on during the Power-Up Phase, or when in Walk-Test Mode.

Control LED On

Control LED Off

PIR LED On/Off Setting

Switch 9 sets if the PIR LED (Green LED) is to be used or not.

PIR LED On

PIR LED Off
Note: When changing switch setting you must re-boot your PixController board. When re-booting you must wait approximately 30 seconds before turning power on again. Not doing so can result in the controller not working properly. Symptoms of this are a dim red LED or blinking green LED, or both.
Copyright , PixController, Inc. http://www.pixcontroller.com, all rights reserved. PixController, Inc. 2610 Haymaker Farm Road Export, PA 15632