The throwaway camera established a new market niche at the low end of the consumer market. Companies such as Kodak (Rochester, N.Y.) have found a way to pare down the elements of a camera to packaging material and film. A few plastic parts, the shutter and lens, complete the design, which competes quite well in terms of image quality with more expensive products. The key ingredients are a reasonable-quality plastic lens and of course the film, which is always high resolution.
Kodak has perfected the manufacture of high-quality plastic lenses. The technique involves more than close tolerances in shaping the lens. With traditional glass optics, the glass itself provides the quality medium for focusing light. The task is to find an efficient way to shape the glass for optimal performance.
With plastic, those physical parameters are reversed. Molds can be used which solve the shape problem easily. The determinant of quality becomes the plastic itself-and quality depends on the time set aside for curing a given volume of plastic. A short curing time will result in refractive index variations within the volume of plastic that will distort the ability of the lens to focus light.
Thus, better plastic lenses will result from a longer curing time, but that in turn drives up the cost, which is sensitive to manufacturing throughput.
We have tackled the problem of cheap lenses from an entirely different perspective. By increasing the image-processing capability of the electronic portion of the imaging system, it is possible to reduce many of the physical limitations of a camera. Physical optics have been improving at a linear rate, while electronic sensor and image-processing chips are improving at an exponential rate. It therefore makes sense to put engineering effort into improving the electronic portion of a system. Given the different rates of technology advance between physics and electronics, the potential payoff is much greater.
Specifically, we have devised a simple approach to plastic lens manufacturing. Since all molds are essentially equal in terms of cost of production, we have come up with a micron-level lens shape that encodes optical wavefronts accurately. Coupled with the unique lens geometry, specially designed image-processing algorithms are able to extract an accurate image from the data arriving at an image sensor behind the lens. The "image" picked up by the sensor would be impossible for the eye to resolve, yet it contains all the information in the original object being imaged.
Delivering the image in this form and then processing it in software provides some additional cost saving. Many parameters such as depth of field can be varied in software.
To deliver this capability in a low-cost handheld consumer device, a DSP that could execute these compute-intensive algorithms with minimal power would be required. We are working with Oxford Micro Devices Inc. (Monroe, Conn.) Ax36 video DSP chips for this reason. The architecture is flexibly programmable without sacrificing raw throughput.
