Imaging (now digital)

The stately history of analog photography  Leonardo da Vinci – cameras obscura – 1519  Photosensitive materials explored – 16xx – 1830 (and beyond)  Daguerrotype – 1839  Wet Plate processes – 1860 First impact on public awareness of the wider world  Color photography – 1868 (first plausible process)  Small cameras (Leica, Ermanox, Rolleiflex) 1914 – 1920s  Japanese entry into camera business – 1950s  Automation Auto exposure – Auto focus – 1980  Film phaseout in favor of digital – 1990 to 2010

The Pixel Race  The roots of digital imaging – 1960s and 1970s CCD and CMOS photosensors first developed (1969 CCD patent) Cold War spy satellites needed to eliminate dropping film by parachute! CMOS – cheaper to manufacture, lower basic quality, able to add image processing on the same chip to compensate CCD – nonstandard process, more expensive (and often uses outdated facilities) higher quality, but unable to integrate  Quickly dominated optical astronomy  The transition in cameras for terrestrial use 1985 introduced digital camera backs ($100K) Mid 1990’s – professional and consumer cameras Present status (e.g. Olympus)  Since 1950s still supports 22 film cameras, sells about 8 more  Since 1996 still supports 46 digital cameras, selling 40 more

What’s inside a digital camera  Simplest case: tethered camera Fixed focus (simple lens in a threaded mount) Two chips

How to capture colors

Standard (Bayer) imager layout

Nonstandard Imagers  Novel layouts (Fuji) Conventional array has greatest resolution along the 45 degree axis. Why not tilt the array to maximize x and y resolution instead…? Fuji is also exploring split cells to enhance dynamic range  Novel 3-color single cells (Foveon)

We’re not done yet – other tasks  Automatic exposure  Automatic focus Active – passive  Passive is dominant for digital Most common scheme is contrast enhancement, based on the actual image Better is phase sensitive detection, since it gives direction in which to correct

How big is digital film?  Digital backs support astronomical pixel arrays ($20-50K) Mpixels and 60 x 45 mm substrate  Highest quality digital SLRs (Nikon, Canon at $8-10K) 8, 12, 16 Mpixels and 24 x 35 mm substrate “full frame” 35 mm  New “low end” DSLRs (Nikon D70, Canon 20D, Olympus E-1) 5-8 Mpixels, “magnification factor of 1.5 (APS) to 2 (4/3)  Consumer digicams (use 2/3, 1/1.8, … tiny chips) “prosumer” 5-8 Mpixels, non-interchangeable lens < $1000 “consumer” 3-5 Mpixels, point and shoot <$400  Lenses must shrink to match imager sizes

Side Effects in the MPixel Race  Using whole cell for imaging gives best dynamic range, lowest noise, greatest sensitivity This sacrifices video output (used in LCD viewers)  Drawbacks of fractional cells compensated by microlenses  Lens design for smaller image areas is different Short back focus Need to restrict angle of light Wide angles become normal

Where is(was?) the quality crossover?  Goto Norm Koren’s discussion  

Where can we go with this?  What type of image taking is a digital camera not good for? Why?  The relationship between image resolution and information Face recognition Text recognition Object recognition  Develop a rough spec for a camera that does one of the following ON-BOARD Text translation (OCR) Continuous wireless upload Wireless printing Gesture learning and recognition (for shoot still, shoot video, stop video, etc) Motion detection and alarm Face learning and recognition Object learning and recognition Visual temperature detection and alarm NOW let’s make the camera we need for our application:  Given a power supply, an imaging chip, and flash memory, what would you need to do to make your own digital camera? What additional parts will you need?