Rochester Institute of Technology RIT Scholar Works Theses Thesis/Dissertation Collections 5-1-1978 Image Quality of a Multiple-Pass Optical Contrast Enhancement Technique Dudley Boden Follow this and additional works at: http://scholarworks.rit.edu/theses Recommended Citation Boden, Dudley, "Image Quality of a Multiple-Pass Optical Contrast Enhancement Technique" (1978). Thesis. Rochester Institute of Technology. Accessed from This Thesis is brought to you for free and open access by the Thesis/Dissertation Collections at RIT Scholar Works. It has been accepted for inclusion in Theses by an authorized administrator of RIT Scholar Works. For more information, please contact ritscholarworks@rit.edu.
IMAGE QUALITY OF A MULTIPLE-PASS OPTICAL CONTRAST ENHANCEMENT TECHNIQUE by Dudley M. Boden A thesis s\ibmitted in partial fulfillment of the requirements for the Bachelor of Science Degree in Photographic Science and Instrumentation at the Rochester Institute of Technology, Rochester, New York May, 1978 Thesis Advisor: Professor John Carson
6 <\V\^ ABSTRACT People have been looking for means of recovering under developed images since the inception of photography. The use of two partially reflective parallel mirrors on either side of the negative is one of the more simple possabilities. This system has been investigated to see how much contrast enhancement it provides and what affect it has on resolution and graininess. It has been found that contrast gains of 1.5 to about 3.0 times can be expected for images with densities from.1.4 to.30. Also the resolution of low contrast images is increased slightly and graininess is decreased by almost three times. However, the effectiveness of the system does seem to be limited to very low density images.
TABLE OF CONTENTS I. Introduction Page 1 II. Experimental Page h III. Results Paee 9 IV. Discussion Page 13 V. Conclusion Page 16
INTRODUCTION Methods of recovering images that have been lost to either underexposure or underdevelopment have been the subject of much research throughout photographic history. The problem is to find a way of increasing effective speed or contrast in an already developed image without significantly altering the other characteristics of the film. One approach to this problem has been to pass the illuminating light through the transparency more than one time, thereby increasing the effective absorption of low density areas. The multiple passing of light through a transparency has been accomplished many ways. One of the least complicated methods that has been tested is the direct deposition of a silver coating onto both sides of a negative. This method is, however, still rather complicated and demands special equipment for the deposition of the silver. If this reflective coating could be placed on glass surfaces and then simply brought into contact with the transparency this -would greatly increase the simplicity of the system.
-2- Theory ; The conditions which exist when the mirrors are brought into contact with the film are illustrated in figure I. mirror surfaces Fig. I In the figure, t is the transmittance of the mirrors, t,. is m f the transmittance of the film, and R is the reflectance of the mirrors. From the illustration it can be seen that the resultant transmittance of the film-mirror sandwich is given 2 2 2 4 4 6 6 by the summation; t t, (1 + tv R + tc R + t,. R + J m f f f f ) 2 which approaches cir the limit t t, m f 1-2 RZ2 tzf
The contrast gain; here defined as the ratio of the contrast with the mirrors to the contrast without the mirrors, is then given by 2 the equation, 2. (2) 1-2 2 trf R From this equation it can be seen that large contrast gains should be obtainable in low density images, with mirrors of high reflectance. Objectives: (1) To determine the contrast increase in a system utilizing two partially reflective mirrors. (2) To determine the effect of this system on resolution and graininess. -3-
' "" rrr- -4- EXPERIMENTAL The mirrors acquired for use in this research were dichroic mirrors with reflectance of approximately 97% between 4700 and 6700A. Beyond these points there was a slight peak in transmittance giving the mirrors an overall magenta cast. The coatings were placed on micro-glass which was approximately 2 millimeters thick. This substrate was chosen in order to minimize image degradation due to the thickness of the glass. The transmittance curve for the mirrors is given in figure II. The first set of tests run on this system were tests of contrast gain. 1 - I *+00 500 600 700 Fig. II spectral transmittance of over visible mirrors, range.
For this part of the experiment three films were used; Kodak xrp x-ray film, Kodak Plus x pan, and Kodak Plus x professional. Each of these films was exposed in a Kodak 101 sensitometer with a 2.1 neutral density filter and a number 2 step tablet. Each film was then developed; the Elus x films in D76 and the x-ray film in Kodak Liquid x-ray developer. Each was developed according to instructions except the Plus x professional which was underdeveloped by half a minute in order to slightly extend the toe of the curve. After develop ment, each film was then measured for visual density on a Macbeth TD102 densitometer. The sample size used throughout the experiment was five. Characteristic curves for each sample were then plotted using relative log exposure. A point was then chosen on the curves which was.2 log exposure units greater than the lowest exposure level. At this point the contrast was determined for each sample and the mean and standard deviation calculated for each film. Next, this procedure was repeated for each film, with the mirrors being held in contact with the film with the reflective surface on the inside of the sandwich. A test of hypothesis was then run to determine whether there was a statistically significant increase in contrast. The confidence level used for each test was 90%. This procedure was then repeated for the mirror sandwich with the densitometer in. the green position. Then the mirror system was tested with -5-
oil immersion fluid acting as a liquid gate between the mirrors and film. Finally spacers were made out of pieces of paper with holes cut in them, these were then placed between the mirrors and the film and then evaluated for contrast enhancement. After the system had been tested for contrast enhancement under various conditions it was tested for its effect on resolution. For these tests only the plus x film was used. In order to test the effects of the system on the resolution of images in the toe of the curve, it was desired to image low contrast resolution targets. To image resolution targets at the desired contrast, a device was used which allowed the contrast of the target to be varied (3) over a large range. This device was designed by Steven Neek ' and was comprised of two light boxes which projected even illumination through diffuse glass. The light from the two boxes was then superimposed on each other by a series of mirrors. The contrast of a resolution target placed at one of the light boxes can then be controlled by adjusting the intensity of the target light and the flare light. Prior to the use of this system, the actual contrast values given by each setting on the boxes had to be determined. In order to do this, the multibar high contrast target which was to be used was imaged through this system at a high enough magnifi cation to allow readings of the high and low density areas of the target with a conventional densitometer. The film used for this was also plus x. Using the characteristic curve of -6-
the film to determine effective exposure from the density readings, an estimate of the contrast could be made. This was done for a series of settings on the light boxes. Once an appropriate contrast setting had been chosen the target was imaged on the plus x film at a reduction of approximately 8.5. The targets were exposed on the film so that one set was deep into the toe of the curve, one set was slightly higher in the toe, and one set was centered around a density of 1.0. The evaluation of these images was done visually using a microscope. In order to hold the mirrors in good contact with the film during this evaluation, a set of holders was machined for the mirrors. Each holder consisted of a piece of thin plexiglass about two inches square. Through the center of the plexiglass a hole was drilled a half inch in diameter, then an area around the hole equal in diameter to the mirrors was etched down to a point where the mirrors stuck above the plexiglass by about 1/5000 of an inch. Using these holders each image was evaluated under each of the conditions being tested; without mirrors, with mirrors, with mirrors and liquid gate, and with mirrors and spaces. '=QraErES==sz=^^ Fig. Ill Holder, mirror, fllm sandwl^h.
The final set of tests run on the system were graininess tests. In order to get some idea of the effect the mirrors had on the graininess of the film, a series of low uniform density patches were exposed on the plus x film. To do this the film was loaded into a 35mm camera from which the lens was removed. This was then placed under an enlarger which provided the even illumination. The density of these patches was approximately 0.6. The patches were then reimaged using a microscope camera with a 4x objective. In this manner, the patches with and without the mirrors could be imaged at the same density. These copies were then projected in a dark room and subjects were asked to move closer to the image until they could just discern the grain. Since these were merely crude estimates of graininess, no extensive statistical analysis was performed on the data. This concluded the experimental phase of the project. -8-
^- T7~ RESULTS The contrast of each of the films was determined at a point.2 log exposure units greater than the lowest incremental exposure, 1.0 I i -".8 * D.6.k I * *.S At > ^ ** *.2 ; ** 6.2 OA 0.6 0.8 1.0 1.2 log'e Fig. IV: Dotted line is negative with mirrors, solid line is film alone Contrast Values Plus x without mirrors with mirrors mirrors and spacers 0.13 0.37 0.34 0.15 0.30 0.18 0.21 0.29 0.21 0.14 0.21 0.27 0.18 0.22 0.29-9-
Plus x Professional without mirrors with mirrors 0.15 0.14 0.13 0.14 0.14 0.30 0.33 0.32 0.31 0.35 XRP : X-ray without mirrors with mirrors 0.18 0.20 0.24 0.19 0.20 0.33 0.37 0.41 0.34 0.37 Average Values normal mirror 90% confidence theoretical contrast contrast contrast interval contrast gain Plus X.162.278.196 -.360.260 1.72 Plux x Prof..140.320.295 -.345.320 2.28 XRP.202.364.324 -.404.396 1.80 The test of hypothesis proved that there was a significant difference between the contrast of the film wi.th and without the mirrors. However, there was no significant difference between the contrast of the film and mirrors and the contrast of the film mirrors and spacers. The contrast of the system with liquid gate was so much lower that no statistical tests were run on these values. -10-
-11- Once the 90% confidence intervals were put on the contrast values with mirrors it was noted that the theoretical contrast fell within those limits. So, to within. the accuracy of the tests performed, the system would have to be considered as functioning to the maximum theoretical limits. Also, the incidental test of the system to green light proved to have no significance. The resolution tests for the plus x film produced the following results. Plus x resolution line-pairs/mm without mirrors with mirrors With mirrors and spaces Very low density 8.77 11.05 resolution low density 13.92 17.55 13.92 Normal density 22.11 27.86 From these results it can be seen that the mirrors increased the effective resolution by one set of bars in the target. The spacers, however, negated this effective increase. The graininess evaluation produced the following results in the number of feet the subjects stood from, the givenj i i projected image of a uniform sample, when the grain was i first' visible.
-12- Graininess in distance of subject from image subject no mirrors mirrors 1 20 7 2 18 5 3 17.5 7 4 18 6 5 19 5.5 6 18.5 6 7 18 8 8 16 5 average 18.12 6.19 No further evaluation of this data was done due to the many difficulties encountered in the procedure, rendering the data questionable. One problem arose in the copying of the images with the mocroscope camera. It became very difficult to keep the image of the grain in focus while holding the mirrors in place. A greater problem arose, however, in the subjective evaluation of the images. The cause of the difficulties was large scale modeling which became apparent when the images were photographed of the mirror sandwich. It is believed that this large scale modeling may have influenced the judgments of the subjects.
DISCUSSION Of the various methods of utilizing the plane mirrors in this research, the method that proved most effective was also the most simple. This is fortunate for it was found to be very awkward and messy to use either liquid gate or spacers, between the mirrors and filter. It is still likely that the liquid gate approach may be more effective when ordinary metallic mirrors are used, however, the liquid apparently destroyed the reflectivity of the dichroic mirrors used in this experiment because they were designed for air incidence. Due to the fact that with the liquid gate neither the contrast nor the resolution significantly increased, this system had no benefits. The mirrors with the spacers gave contrast increases equivalent to, although not in excess of, that attained with just the mirrors. The resolution with this system was, however, not increased over the resolution of the film alone. This means that although large areas that have been underexposed may be made more discernable by this system, the fine detail will still not be improved. The reason for this is most likely because the space between the mirrors and the film allowed "more scattering of light than there was when the mirrors were in contact, thus causing diffusion of the image. -13-
When just the mirrors were used in direct contact with the film, contrast gains of between two and three were obtained and the resolution was increased by approximately 1.26 times. What is unusual is that the resolution gain was the same for the high as well as the low density images. Theoretically, the normal density images should be effected less than the low density images. One possible explanation of this effect is that due to the unspecular nature of the light which was used to illuminate the images, the more times the light passed back through the negative, the more it diffused the image. Therefore, the effect of increased diffusion could counter the effect of increased contrast, in high frequency low density images. This effect would also account for the decreased apparent graininess of low density images when the mirrors are used. Also the fact that the mirrors were never perfectly parallel would also cause increased blurring of low density high frequency images. One way of possibly decreasing the diffusion of light by the system would be to use specular illumination. However, this type of illumination would probably increase another problem which was encountered with this system, that of interference fringes. Often while visually examining a negative in combination with the mirrors, interference fringes would appear in the image. All that was required to remove them 14-
was to slightly reposition the mirrors. However, they were still quite annoying. This was the reason the liquid gate was tried between the mirror and film. The hope was that by eliminating the air interfaces the unwanted reflections at these faces could be eliminated. This, in fact, proved to be the case. The liquid gate, however, also eliminated the contrast enhancement. The spacers, on the other hand, did prove to eliminate the fringes without harming the contrast enhancement. This would, therefore, be the preferable system as long as you were not concerned with increasing the resolu tion of the image. Perhaps the best results could be obtained using either specular illumination and spacers or specular illumination, liquid gate and mirrors that function in the liquid medium. -15-
CONCLUSION This research has shown that parallel partially reflecting mirrors can be used for contrast enhancement of low density, low contrast images. The system that proved most effective overall was that with mirrors in physical contact with the negative. This system has also proven to increase the resolution of low contrast images both low density and normal density by a factor of approximately 1.26. The graininess was shown to be actually decreased by the use of this system. However, the validity of this data is somewhat in doubt. The actual practical value of the system is subject to question. One reason is that the mirrors used for this research were somewhat expensive to be used for normal pictorial reproduction. There are many other systems of contrast enhancement that have attained greater contrast gains. This system does have some merit, however, in that it actually increases the effective resolution and decreases graininess whereas most systems do the exact opposite. The lower the base plus fog density of the transparencies, the more effective the system becomes. In general, it is a system which may have practical applications in very specific cases, but, for general usage it is of very limited value. -16-
FOOTNOTES 1. 0. A. Ullrich and P. G. Andrus, "Simple Contrast Enhancement viewer for Film", J. App. Photogr. Eng., 3: 150-153 (1977) 2. Ibid 3. S. Neek, R.I.T. Senior Research Thesis, Aug.- (1977).
REFERENCES G. T. Bauer, "The Use of Partially Increase the Contrast of Images", Applied Optics, 5: 1361-1364 (1966). Transparent Plates to J. M. Sturge, Neblette's Handbook of Photography 7th Ed., Van Nostrand Reinhold Co. 1977. and Reprography, M. Cloupeau, "The Printing of Underexposed Photographs by Means of Optical Contrasters", Photogr. Sci. Eng., 5: 175-180 (1961). 0. A. Ullrich and P. G. Andrus, "Simple Contrast-Enhancement Viewer for Film", J. App. Photogr. Eng., 3: 150-153 (1977) S. Neek, R.I.T. Senior Research Thesis, Aug. (1977).
ACKNOWLEDGMENTS I would like to thank Mr. Harry Bostely of Eastman Kodak for providing the x ray film and developer used in this experiment. I would especially like to thank Professor John Carson for his assistance and guidance throughout the project.