Résultats d'analyse d'image

The analysis of the negative (and also first generation positive black and white and color prints) included the following steps:

1. linear and angular measurements
2. micro-densitometry scans to establish optical density range,
3. black and white photographic enlargements using papers having different spectral sensitivities, and
4. computer-based contrast enhancements.

Linear and Angular Image Measurements

Fig. 3. Horizontal and vertical angles in the photograph

Fig. 4. Summary of elevation angles

Selected linear measurements of the disc's image on the enlarged print are given earlier (cf. The Photograph and Negative). The angular measurements shown in Figure 3 were determined on the basis of the linear measurements of the image, the camera lens' focal length, and on-site survey which is described below. Referring to Figure 3 it may be pointed out that the photograph's center was elevated about nine degrees arc above the horizontal and the tip of the mountain was very close to the geometric center of the photograph. The disc subtended 1.3 degrees arc. Elevation angles from the local horizontal were measured with a surveyor's transit to the top of the mountain, the location of the photographer, and the sun. These latter measurements were obtained at the same time of day, day of the year, and location as that of the original photograph but two years later. These results are presented in Figure 4 with the disc shown in side view. The exact outline of the disc is not known; also, the dome-like protruberance is not shown here.

Micro-densitometry Scan Results

A Joyce Loebl Recording Micro-densitometer with x20 power objective, 50:1 linear magnification ratio, slit width of 0.02 inches and vertical slit height of 1 mm was used on the original negative.

Fig. 5. Micro-densitometer vertical scan through disc image

Density calibration was carried out using a Kodak ND step wedge spanning the densities on the negative. Greatest optical density (brightest positive image) was approximately 0.65 to 0.7 log₁₀ ND and was found on the sunlit cloud. This is equivalent to about 12,500 ft-L luminance. Figure 5 is a vertical scan through the disc's two brightest areas using the micro-densitometer. The tracing peak marked T represents the upper-most (dome) bright area and B represents the lower area of brightness on the front edge of the disc. This scan line is shown in Figure 6 which is an enlargement of the disc's image. Points T and B both have optical densities of about 0.55 to 0.6.

Fig. 6. Photo enlargement of disc

Optical density of the blue sky on the negative is shown in Figure 5 and has a value of approximately 0.4 log₁₀. The gradual slope of this densitometry tracing is due to the progressive sky brightness increase from the zenith to the horizon while the smaller amplitude deviations are due to single and grouped film grains.

Of particular note is the fact that the brightest area on the disc was of lower brightness than the cloud by approximately 0.15 log₁₀ unit. According to a physics handbook (Allen, 1963), a smooth, polished silver surface reflects (within the visible spectrum) increasingly higher percentages of incident radiation with increasing wavelength. An average reflectance value of about 90 % is found. Polished aluminum reflects about 85 % regardless of wavelength of the incident radiation; this is also true for nickel (reflectance of about 60 %), silicon (about 30 %), and steel (about 54 %). This comparison of dark areas on the negative suggests that the surface of the disc is very likely not a polished surface of any of the above metals. If direct sunlight has a brightness of about 750 000 ft-L and a 90 % reflectance is assumed for the disc's surface, the brightest area should produce a brightness of about 675 000 ft-L which is more than an order of magnitude greater than what was found on the negative.

A horizontal scan using the micro-densitometer also was made to see if there was any evidence of a double exposure. A double exposure might be indicated by the presence of double edges if the film registration is not precisely the same during a manual rewind. No such evidence was found. In addition, this camera could not take double exposures due to its frame locking mechanism.

Black and White Enlargements on Different Wavelength Sensitive Paper

Fig 7. Black and white image enlargement: (a) Panchromatic paper, (b) Blue-green sensitive paper

The disc area of the negative was enlarged and printed on panchromatic film which provided a relatively complete and undistorted translation of the three primary colors in the negative into shades of gray. This is shown in Figure 7(a). The top "dome" protruberance is clearly visible. The same area on the negative also was printed at the same enlargement using blue-green sensitive paper which significantly reduces the contribution of the red emulsion layer to the final black and white print. This is shown in Figure 7(b). The bluegreen sensitive paper increases the overall brightness of the sky and causes the "dome" area on the disc to almost disappear. Apparently, the dome is not reflecting or emitting radiation in the red end of the spectrum.

Computer-Based Contrast Enhancement Results

Fig. 8. Color enhancement of disc image using a blue filter

The negative was digitized using a Perkin-Elmer Scanning Densitometer set to six microns diameter entrance aperture. Optical density was achieved to 16-bit resolution. The negative was scanned and digitized three times; each scan was made using a color filter having the approximate spectral distribution as the dye layers in the negative [The film's yellow forming layer, magenta layer, and cyan layer peaked in sensitivity at 425, 545, and 650 nm, respectively. This technique permitted the information content in each layer to be analyzed separately]. System output was recorded on 1/2 inch magnetic tape at 1,600 bits per inch density for processing and display by a digital computer. The region studied was only slightly larger than that of the disc in order to conserve memory and processing time. The range of optical densities found within this image area ranged from 5 to 400. Since the optical densities extended only from about 200 to 400 for the disc's image, the computational range was truncated by dropping the top 8 bits. Thus, the 8 bits from zero to 255 levels of density are presented in all of the following computer color enhancements.

Figures 8 and 9 are enhancements obtained using only the blue filter scan, that is, there is almost no contribution to this image from green or red wavelengths. A very high contrast color enhancement is shown in Figure 8 using blue picture elements (pixels) on the cathode ray tube monitor to represent film densities associated with the image of the disc and orange and red pixels to represent film densities characteristic of the surrounding sky density. It must be emphasized that there is no particular significance to the colors seen in these computer-enhanced photographs.

Fig. 9 - Black-and-white enhancement with undistorted contrast

Figure 9 shows a black and white enhancement using an undistorted contrast. Much of the top surface detail becomes invisible in both Figures 8 and 9. Presumably this is due to the particular range of wave-lengths that are being reflected or emitted by the disc. Both of these figures show that the sky is relatively homogeneous, the film's crystals (each possessing different sensitivity to light) are approximately random in their spatial distribution, as expected. Also shown is a relatively sharply defined bottom edge of the disc relative to its upper edge. The shadow seen under the disc's lower edge in Figure 6 is barely evident here.

Fig. 10 - Color enhancement of disc image using a green filter

Figure 10 presents an enhancement made using a green filter where purples and blues are assigned to densities which predominate within the image of the disc and yellows are assigned to densities characterizing the background sky. A long vertical scratch exists to the left of the disc's image within this particular emulsion layer. Not only is the sky's density relatively homogeneous in its density but the two regions of greater brightness on the disc become much more apparent. The overall shape of the disc is symmetrical but has more pointed ends. The significance of this is unclear.

A red filter was used to generate the enhancement shown in Figure 11. As was noted in enhancements using a green filter, the dome is missing in this enhancement suggesting that the protruberance on top of the disc is reflecting or emitting wavelengths mainly in the blue-green portion of the spectrum.

Fig. 11 - Color enhancement of disc image using a red filter

Finally, a three-color composite enhancement including blue, green, and red wavelengths was made. Figure 12 presents one such enhancement to illustrate the homogeneity of the sky as well as the emulsion flaw and highlights on the image of the disc. None of the above enhancements show any evidence for a suspension line or thread above the disc.