MegaVision Imaging Products
It is commonly recognized, if not grudgingly admitted, that a single-shot color image capture is a compromise of convenience. While a single shot image capture is indeed convenient, it is fraught with compromise that often results in less than perfect quality as well as increased overall production costs, particularly if high image quality and good color fidelity are desired. Such problems as filter factor differences, under-sampling color and luminance artifacts of decaled (such as Bayer pattern) sensors, and filter mismatch to CIE color space are well known.
By capturing multiple images of a scene, the problems inherent in single-shot image capture may be overcome. The provision is, of course, that both the scene and the camera must remain stationery throughout the image capture.
N-Shot Multi-Spectral Capture
The professional photographer who demands perfection knows very well how much time is spent performing color corrections. No matter how well the digital camera is profiled and color managed, localized color correction is necessary to achieve the level of color matching required for many commercial, artistic, and academic goals. The reason for this is simple and fundamental: No three color filters exist which accurately model the human visual tristimulus color response. Thus, while certain colors may capture and render accurately, it is impossible to accurately capture colors over a broad range of commonly encountered dyes, pigments, paints, fabrics, materials, papers, and other reflective surfaces. While any color can be made sufficiently accurate within the bounds of available reproduction systems, the post-processing required to achieve the desired color matching is time-consuming, tedious, and demanding.
MegaVision has developed a revolutionary approach to solve this previously intractable problem. While solving this completely intractable problem, the approach also simplifies other difficulties, improves image quality, and results in more robust system design.
The approach is beguilingly simple. The vast majority of color image capture requires illuminating the scene with nominally white light, then separating the white light into it’s color components after the light is reflected from the scene and before the image is recorded. This separation is usually performed by color filters, though it may be performed by a prism or by the detector itself. Regardless, deficiencies and compromises are unavoidable.
Suppose, for a moment, that the light could be separated into color components BEFORE it illuminated the scene. Imagine removing the color wheel from a triple-shot camera and attaching the color filter wheel to the lights. It would be possible, then, to capture a multi-shot color image by capturing sequential images with the scene being illuminated by a different color of light in each capture. This approach enjoys all the advantages of a triple-shot color wheel image capture (monochrome sensor, full color at every pixel, ideal filter matching, etc.). But putting a color filter wheel in the lights has a number of practical implementation problems. Fortunately, colored light is available from other sources, most notably and practically, LED’s.
What is even more notable is that LED’s are available in a range different colors with nearly ideal spectral properties to illuminate scenes for color image capture that solves the fundamental intractable problem. LED’s may be fabricated to produce rather narrow bands of colored light. By changing LED properties, different LED’s can produce very well specified, very repeatable, and very stable colors that span the visible range. The solution to the intractable problem, then is very simple: replace 3 broad-band color filters with 5, 6, or even 7 different narrow band LED’s. By using multi-spectral, narrow-band illuminants, the rendered colors (transform to CIE color space) match human visual color perception with unprecedented accuracy.
There are additional benefits to this revolutionary approach:
1. Because color filters are eliminated from the path between the scene and the sensor, there is no filter distortion or required compensation, there is one less optical element to diminish the light, there are no filter surface deficiencies (scratches, dust, and reflections) to degrade the image, and there are no filter properties that decay over time.
2. All the light that is put onto the scene is available to create an image. Since there is no filter loss, the light is more efficiently used. Filters used with white light illumination discard most of the available light. Using colored light as the illuminant, there are no unwanted colors to discard. Additionally, because visible light LED’s produce no infra-red, the IR filter of the sensor may also be eliminated, further increasing efficient use of light.
3. LED’s are very efficient, use very little power, and stay cool.
4. Between the elimination of filter loss and the inherent efficiency of LED’s, power consumption is at least an order of magnitude less than for traditional lighting.
5. Because there is no change of filter elements between exposures, image registration for the different color exposures is as perfect as the lens will allow. For highly corrected apochromatic state-of-the-art optics, the images registration is perfect.
6. Because the amount of light may be adjusted and set for each different exposure, the S/N of all exposures is optimized. “Blue channel noise” due to low blue filter efficiency and low detector sensitivity is non-existent.
7. Unlike piezo stepped arrays, there is no possibility of color-decal (Bayer) pattern induced artifacts.
Another Option Is MegaVision's Triple-Shot Color Wheel
The first professional quality color digital images were created by sequentially capturing 3 monochrome images through 3 color filters. MegaVision has reprised this concept with a new USB controlled filter wheel.