Monday, July 27, 2009

Optical-Mechanical Linear Cameras



Conventional cameras focus an image onto a 2-dimensional image sensor. One problem with this is the limitation of resolution imposed by image sensor technology. It is easier to build a 1-dimensional camera and allow the orbital motion of the spacecraft to sweep it across the planet. An innovation often attributed to Landsat-1, Soviet scientists first deployed linear cameras a year earlier, on Luna-19. Built by Arnold Selivanov and Iuri Gektin, they represent an evolution of the panoramic camera used on Luna-9 in 1966.

These cameras, for 1971 and 1974 low-orbit survey, were designed to produce long, high-quality panoramas of the lunar surface. They used a photomultiplier tube (4) as the detector, with a spinning prism to scan a 180° "cylindrical fisheye" image. The scan rate was 4 lines per second. From an altitude of 100 kilometers, the craft could resolve 100 meters along the direction of scanning, and 400 meters along the perpendicular direction of flight. The images extend to the lunar horizon, which was used to help calculate the precise orbital motion of the satellite

The Luna-19 and Luna-22 "heavy orbiters" are still somewhat mysterious missions, although one objective was the mapping of the Moon's uneven gravitational field. Luna-22 adjusted its orbit until it was skimming the lunar surface at 15 to 30 kilometers distance. By one report, Luna-19 returned 5 panoramas and Luna-22 returned 10.


The Mars-4, Mars-5 and Venera-9 orbiters contained linear cameras designed by Gektin and his team. They scanned images 30° wide and arbitrarily long, as the orbit of the spacecraft swept across the planet. The camera design was similar to the cycloramic camera on Luna-9, but its scanning mirror oscillated without the need of a rotating assembly, using the satellite's orbital motion to sweep out an image swath. It used automatic gain control and operated in a logarithmic-photometer mode. Each scanline included some black and white calibration stripes transmitted during the return stroke.

The box, above left, is an analog 4-track tape-loop recording device designed to work with this linear camera. It recorded up to 45 minutes of two 1000 Hz video signals as well as two synchronization signals from the onboard crystal oscillator. Both cameras could be simultaneously recorded for 45 minutes, or one camera could record for 90 minutes. The video could be read and digitized for transmission to Earth, at two speeds (i.e., at two pixels/line resolution).

Reports claim the tape recorder was also used to store the video signal from the lander, although technical papers stress that the radio signal from the Venera and Mars landers to the orbiter was digital, not analog.

The Mars cameras used two photomultiplier tubes and returned images in three wavelength ranges. A PMT-112 (AgOCs cathode) with a red glass long-pass filter was used to image in infrared. A PMT-114 (multialkali cathode, also used on Venera lander) was used with red and orange glass filters to image those colors. The cameras scanned at 4 lines/second, generating 1000 Hz video (250 cycles/line), which was recorded on magnetic tape. The primary readout rate was 1 line/second, transmitted to Earth probably at 256 or 512 pixels/line. The option existed to scan at 4 lines/second and send reduced resolution at higher speed. Mars-4 returned 2 panoramas, and Mars-5 returned 5 panoramas.

The Venus cameras both used the PMT-114 with violet and ultraviolet filters to obtain images in those spectral ranges. It scanned at 2 lines/second, generating 1000 Hz video (500 cycles/line). During transmission to Earth, the tape could be read and transmitted at 256 pixels/line in the primary mode, or at a slower special rate of 512 pixels/line. Venera-9 performed 17 survey missions from October 26 to December 25, 1975, using the ultraviolet camera with the violet camera sometimes recording simultaneously. Resolution was 6.5 to 30 km, depending on the spacecraft altitude.

The panoramas, recorded over 30 to 50 minutes, were probably about 256 × 6000 × 6-bits in size, and contained highly elongated images of the planet. They were contrast enhanced and linearly compressed by scanline averaging, to reduce noise and geometric distortion. These images were higher resolution than the later Pioneer Venus cloud photometer, but unfortunately the images from this survey have never been released to the public. The poor-quality images above are scanned photocopies of printed pictures

In 1988, the Soviet Union launched Fobos-1 and -2, Mars orbiters with small vehicles intended to land on Phobos. Selivanov and Gektin's team designed a 28 kilogram optico-mechanical camera, similar in basic design to the Mars-5/Venera-9 linear cameras. Called TERMOSKAN, the camera contained two detectors: One for 600-950 nm returned images in the red and near-infrared range. The other, cooled by liquid nitrogen, imaged the thermal infrared wavelengths from 8.5 to 12 μm. Seen above is the third of four scans around the equator of Mars, 512×3100 pixels, from Olympus Mons to the Valles Marineris



The spacecraft was 3-axis stabilized, with the TERMOSKAN camera pointed away from the Sun. A moving mirror scanned one dimension at 512 pixels/line and 1 line/second. The nearly circular orbit of the spacecraft moved the camera in a swath across the illuminated face of the planet. The faint horizontal streak is the shadow of Phobos, following the spacecraft's orbit.

Above is a full sized section from the second scan in the far infrared. With 1.8 km resolution, the Fobos-2 images are several times higher resolution than the recent thermal IR images from Mars Global Surveyor. Each scan line consists of 384 pixels of image and 128 pixels of calibration data (which has been omitted). A later version of the camera was installed on Mars-96, which was destroyed in a launch mishap

Linear optical-mechanical cameras have been applied to non-military Earth observation satellites. In the early 1970s, two scanners were developed by Selivanov's team, for the Meteor weather satellites: MSU-M scanned 4 lines/sec by oscillating mirror (similar to the Mars-5 camera). It swept a 3000 km swath at four bands in the visible and infrared. MSU-S scanned 48 lines/sec by spinning prism (similar to the Luna-19 camera). It swept out a 2000 km swath with 240 meter resolution, in two spectral bands.

The two images above show images gathered from MIR in the 1990s. The latest spinning-prism scanner, the MSU-SK, has been installed on Meteor-3M, Okean and Resurs-O satellites, as well as the MIR space station. It sweeps out a 600 km wide swath with an arc-shaped scan, returning up to 4756 pixels/line. It is combined with the MSU-E push-broom camera, which uses three 2048-element linear CCD sensors. The MSU-E returns 200 lines/sec in a 45 - 78 km swath, running down the center of the MSU-SK image. A 24-bit image is returned, consisting of three channels selected from the set of 5 spectral bands on the MSU-SK and 3 bands on MSU-E.



0 comments: