ABSTRACT: A Kodak DCS460 digital camera (with 1300 x 1000, 9-micron square pixels) and a PixelVision, back-illuminated digital CCD camera (with 1024 x 1024 elements, 12-micron square pixels) were tested at calibrated light levels ranging from 10^-2 footcandles to 10^-5 footcandles. The PixelVision back-illuminated CCD was shown to be 10,000 more sensitive than the Kodak DCS460 CCD camera. Surprisingly, the 24-micron PixelVision CCD was shown to have comparable MTF performance to the 9-micron Kodak CCD.

Introduction

Two digital cameras were tested using a calibrated Optoliner^TM uniform light source and an RCA television target. The light source was varied from 10^-2 footcandles to 10^-5 footcandles. As the response of the Kodak DCS460 camera was weak under all of these lighting conditions, a 1-second integration time was used for these measurements. A 1/30 second integration time, much more compatible with digital photography. These images are shown in Figure 1 and Figure 2 for the Kodak DCS460 and the PixelVision, Inc. back-illuminated HIRBI CCD respectively. As can be seen in the images, the sensitivity of the Kodak DCS460 is limited to about 10^-3 footcandles (1-second integration time). At an integration time practical for digital photography, this would correspond to light level of approximately 2 * 10^-2 footcandles. This light level corresponds to dusk conditions.

In comparison, the PixelVision back-illuminated CCD camera was shown to be 10,000 times more sensitive. The camera responded well down below 1 * 10^-5 footcandles. These tests, correspond well with previous tests by the US Army which show that the PixelVision camera has response at below 1 * 10^-6 footcandles.

Shown in Figure 3 is a summary of the data obtained from the images. The last line of the table titled "Contrast" corresponds to the term used by camera manufacturers to calculate signal to noise. The calculation calculates the mean of the "white" signal and divides by the standard deviation of the areas in the dark area of the image. This method neglects the shot noise on the signal, and is actually a measurement of measurable contrast.

Figure 4 plots the signal-to-noise of the two camera types. As can be seen in the graph, the PixelVision back-illuminated CCD had dramatically superior signal to noise over all light levels.

Figure 5, shows a graph of the CTF measured for both cameras. As can be seen, the two sensors have comparable MTF. Taking a column plot through the various spatial pattern targets in the image were used to make the measurements for various spatial resolutions. For the Phase II NASA Johnson SBIR, PixelVision proposes to use 12-micron pixels, which will extend the resolution of the camera out to 45 lp/mm.

In a calibrated test of the Kodak DCS460 and a PixelVision, Inc. back-illuminated CCD, the PixelVision camera was shown to have 10,000 times greater sensitivity. An ISO rating of 80,000 was measured in comparison to the ISO rating of 80 for the Kodak DCS460. Considering, that the PixelVision, Inc. CCD has greater than 16-bits dynamic range (600,000 electrons full well) as compared to the 8-bits dynamic range of the Kodak CCD (35,000 electrons full well) and has comparable MTF and limiting resolution, the PixelVision Inc. camera has greater than four orders of magnitude greater information bandwidth per image, than does the Kodak DCS460.

These tests support the PixelVision approach to providing a multi-CCD high-resolution (4096 x 4096) back-illuminated CCD to meet NASA digital photography needs.