ALDENP. COLVOCORESSES U. S. Geological Survey McLean, Virginia 22101

Image Resolutions for ERTS, and GEMINI/

ARLY IN 1972 the first Resource ERTS-A RBV RESOLUTION E Technology Satellite (ERTS-A)is sched- Image quality tests on the ERTS RBV'S have uled for launch in near-polar orbit. It will been conducted by RCA and accepted by carry three return-beam-vidicon (RBV) TV NASA. The results are contained in Table 1 of cameras and a multispectral scanner (MSS). NASA'sMemo Change #28 to "Opportunities In 1973 a post-Apollo manned space flight for Participation in Space Flight Investiga- called SKYLAB will orbit the earth at an in- tions" dated June 2, 1970. This table uses the clination of 50° to the Equator. In addition to terms resoloing power and ground resolution other sensors it will carry a battery of six (feet/TV line), but neither of these refer to multispectral cameras identified as experi- the photographic criterion of resolution al- ment S190. though they may be indicative of the size of This paper compares the images expected single objects of average contrast and shape from ERTS and SKYLAB with those already ob- which could be detected. However the figures tained from GEMINI/APOLLO, all in terms of given for the 0.690-0.830 pm band (near-IR) the photographic criterion of resolution. Re- are better than those indicated by ERTS spec- cently provided data have led to several ifications (NASA,1970) and by recent tests changes in the resolution figures of ERTS-A (Weinstein, 1971) and therefore questionable. forecast a year ago (Colvocoresses, 1970). The table does indicate how resolution varies with contrast for selected earth surface features. Resolution as used herein refers to the pho- The electronics community describes TV- tographic criterion of image quality as related system resolution in terms of the effective to the observable minimum spacing of bar number of TV lines on the tube (raster). In targets of specified design (Mil-Std 150A, this context, Nordberg of NASA/Goddard 1959). The term is derived from the optical (1970) has estimated that for most ground criterion that defines how close two point- targets, which are of low contrast, the effec- sources of light can be to each other and still tive resolution expressed in terms of TV lines be distinguished as two points. On the other will be below 3,000 for the green and red hand a single object of sufficient brightness bands. This compares to 4,100 (or 4,500) lines but of no real size at all may be detected in a for a high-contrast target, which are figures given image. Stars are the prime example. A quoted for the RBV resolution in TV terminol- linear highly reflective terrestrial object, such ogy and which represent the actual number of as a highway with width as small as one-tenth TV lines (4,100) or the equivalent number of of the resolution figure, may often be de- lines (elements) along a line (4,500) as re- tected on space and aircraft images, but this ported by RCA and NASA. If one takes a mean should not be equated to resolution. In such of 4,100 and 4,500 (4,300) and compares it instances contrast is a decisive factor, and a with 3,000, an inverse ratio between high- and bright (highly reflective) object on a dark low-contrast target resolution is established. background normally has higher detectability Kingslake (1965) and Jensen (1968) agree than a dark object on a bright background. that it takes 242=2.83~2.8 TV lines to re- This is due to the spread function, which is a solve (separate) high-contrast bar targets. characteristic of image formation in both film Thus it takes (4,300/3,000)2.8 or 4 actual TV and TV systems (Schade, 1964). lines to resolve (separate) the low-contrast targets. Publication authorized by the Director, U. S. The rationale of this approach is based on Geological Survey. the vertical resolution where horizontal tar- PHOTOGRAMMETRIC ENGINEERING, 1972

TABLE1. DESIGNSPECIFICATION MULTI-SPECTRAL TV CAMERASYSTEM KESOLVING POWER FOR TYPICALSCENE OBJECTS AND LIGHTING

Average Ground Resolution (Feet/T V Line)

Band (Micrometers) 0.475-0.575 Targets: High Contrast (Laboratory) 150 Desert Sand vs Shadow 165 Average Plant vs Wet Loam m Average Plant vs Dry Loam 180 Average Plant vs Water 275 Average Plant vs Shadow 200

Note by A.P.C. These figures are not those associated with the photographic criterion of resolution and therefore should not be equated to resolution figures quoted for photographic systems such as those flown on GEMINI/APOLLO or defined for SKYLAB. gets (bars) lying parallel to the TV lines must an altitude of 435 km, and the S190 cameras, be resolved. This is considered to be the limit- with a focal length of 152 mm. (6 inches) will, ing condition, and it has been assumed that according to ITEK, resolve from 29 to 138 lines horizontal resolution would be designed to per mm. on the film. ITEK is using photo- match that of the vertical. Recent tests (Weinstein, 1971) indicate that horizontal from 10 meters for a high-contrast target on high- resolution will be better than the vertical. resolution film to 40 meters for a low-contrast tar- However, until RBV imagery has been fully get on low-resolution film. The typical low-con- tested to resolve bar targets under varying trast earth scene will probably average about 25 meters resolution (ground) for the various films conditions, Weinstein's most recent figures planned. The normalized resolution figure (ground) are not considered appropriate to compare for this Hycon camera, as flown in SKYLAB,and with photographic systems. The contrast ra- compared to GEMINI/APOLLO photography, is pre- tios of 1,000:l and 1.6:l have been arbi- dicted as 0.25. That is, the Hycon camera will re- solve (separate) objects of one-fourth the size of trarily assigned, but they are commonly those resolved on typical GEMINI/APOLLO photos. accepted to define high and low contrast. Detection of single objects will probably bear a These must of course be measured at the ap- similar relationship. (Hycon is now Actron). propriate wavelengths. The 4,100 TV lines re- cord on a 25.4-mm (1-inch) raster (format) TABLE2. EXPECTEDERTS-A, RBV and the focal length of the camerasis 125 mm. RESOLUTION(PHOTOGRAPHIC (-5 inches). At the designated altitude of CRITERION)'GREEN AND 913 km, a TV linewidth scales to 45 meters on RED BANDS^ the ground. Thus resolution in terms of dis- tances on the ground is given in Table 2. Noii$TVObject pair sparing on ground ERTS-A MSS RESOLUTION Object type rewired to The ERTS MSS will have somewhat lower resolve (sep- resolution as the instantaneous spot size is de- arate) an Meters Feet object pair fined as 79 m (260 ft.) on the ground. Equat- ing spot size to TV lines and converting to the High (1,000: 1) photographic criterion results in resolution contrast 2.8 126 415 (ground) figures of 224 m at high contrast and Low (1.6: 1) 316 rn at low contrast for the MSS. contrast 4.0 180 590

SKYLAB(S190) RESOLUTION* Figures are based on an optimum image-record- corporation (1971) issued some ing system; image processing will involve some deg- lution predictions for ~190. fly at radation as will image motion which is about 75 sKYLAB m (ground) in the direction of flight at the 12 ms * During June 1971, after submission of this paper for publication, NASA authorized the inclu- ' The third camera (near-*R based On of the Hycon 18-inch focal length frame cam- recent tests (Weinstein, 1971) has a calculated era (4.5 by 4.5-inch format) for SKnAB. Based on resolution (ground) of 156 m at high contrast and Hycon estimates, the resolution (ground) will vary 275 meters at low contrast. IMAGE RESOLUTION FROM SATELLITES

TABLE3. RESOLUTIONCOMPARISON, PHOTOGRAPHIC CRITERION (meters on the ground)

E TTS-A Skylab Sl9O Gemimi/Apollo - RBV bands Hi. Low Hi. Low MSS res. res. res. res. Green Red IR film jilm jilm jilm

High contrast 1000: 1 126 126 156 224 22 49 so* 80* Low contrast 1.6: 1 180 180 275 316 38 99 70 125

- - * Slater uses 100:l rather than 1000: 1 to define high contrast, but the differences are not significant. graphic terminology, and their figures ac- data available and are applicable to typical tually refer to line pairs per mm. (Ip/mm.). earth scenes of low contrast. They indicate At high (1,000:l) contrast, the predicted that objects to be resolved (separated) on resolution (ground) will vary from 22 to 49 m. ERTS imag~rymust be of at least twice (2.2) At the expected scene contrasts of 1:6 to 1, the dimension of those resolved on GEMINI/ the resolution (ground) will vary from 38 to APOLLO,whereas SKYLAB will resolve objects 99 m., depending on the camera and film considerably smaller (0.7 of GEMINI/APOLLO). combinations used in the 6-camera system. Detection of single objects will probably bear a similar relationship.

Several good Gemini and Apollo photo- graphs have been subjected to resolution analysis by Data Corp. (1965), University of 1. Colvocoresses, "ERTS-A Satellite Imagery," Arizona (Slater 1969), RCA, and others. The Photogrammetric Engineering, June 1970. 2. "Military Standard Photographic Lenses," published resolution figures (Slater) range (Mil-Std-150A) U. S. Govt. Printing Office, from 50 m for a high-contrast object on high- 19.59) resolution film to 125 m. for low-contrast ob- 3. SchLde, "An Evaluation of Photographic jects on low-resolution film. Figures given are Image Quality and Resolving Power," Journal of the SMPTE, pb. 1964. ground distances based on original film and 4. NASA Goddard, Design Study Speci,f$ations ideal object-viewing and image-recording con- for the Earth Resource Technology, S-701- ditions from a nominal altitude of 230 km. P-3 Rev. B, Jan. 1970. 5. Weinstein, Miller, and Barletta, "Simulation (125 naut. mi.). of ERTS RBV Imagery," paper presented at 7th International Symposium of Remote Sens- ing of Environment, University of Michigan. The photographic criterion of resolution is ~i~1971. recognized as an arbitrary, and somewhat 6. Nordberg, William, memorandum of April 14, 1970. Subject: "Minutes of ERTS User Agency limited, measure of image quality. It is, how- Meetinq," March 13, 1970. ever, quantitative, understandable, in com- 7. Kingslake, Applied Optics and Optical Engi- mon use, and relatively objective. It is taken neering. Academic Press, Vol. 11, 1965. as the most logical basis for a simple compaii- 8. Jensen, Optical and Photographic Reconnais- sance Systems. John Wiley and Sons, 1968. son of specific TV, scanner, and photographic 9. ITEK Technical Bulletin, ITEK Mziltispectral systems as indicated in Table 312. Photographzc Faczlity. ITEK 1971. From this table, normalized resolution fig- 10. Data Corp., Edge Analysis of Color Photography. Prepared for NASA Headquar- ures for ERTS and SKYLAB as related to ters, 1965. GEMINI/APOLLU are derived as follows: 11. Slater, Keenan, "Preliminary Post-Fight Cali- bration on Multiband Photography Experiment S065." Optical Sciences Center, University of Arizofa, Sept. 1969. 12. Rosenberg, Paul, Resolution, Detectability and Recognizability," Photogrammetric En- These figures are based on the limited test gineering, December 1971.