Achieving high-resolution imagery of distant terrestrial objects from ground-based imaging sensors presents a unique technical challenge. The entire optical path is fully immersed in a dense and turbulent atmosphere, resulting in a significant loss of scene contrast and angular resolution. An uncompensated atmosphere is generally the dominant system limit to angular resolution for telescope apertures larger than a few centimeters.
Different technologies including adaptive optics, phase diversity imaging, and various forms of speckle imaging have been used— often in combination— to compensate for atmospheric turbulence blurring. The Miniature Integrated Speckle imaging Telescope (MIST) offers distinct benefits in this area.
Long-range, high-resolution speckle imaging system requirements are conceptually simple: a well chosen seeing environment and observing site, a good quality telescope coupled to a short exposure sequence camera, and a suitable image processor. The environment includes the scene illumination characteristics, the weather, atmospheric seeing and attenuation along the entire optical path, and selection of the optimal observing site. The telescopic sensor assembly includes the telescope, filters, sensors, image preprocessors, image recorders, and the mechanical skeletal framework. The image processor provides contrast enhancement, resolution enhancement, and image fusion.
For a quasi-stationary scene, an image sequence is recorded. There are a variety of speckle-image-processing techniques varying from simple image sorting 1 to bispectrum and multiframe blind deconvolutions 2,3. The seeing compensated images here are formed through a four-step image processing workflow:
Figure 1. These images of a radome (radar dome) at a distance of 128.3km (79.7 miles) demonstrate time-varying atmospheric lensing along the line of sight. Five of the photographs were acquired within a one-minute interval. The lower left image was acquired about 10 minutes after the atmosphere had settled down a bit. A zenith view to space and a 10km horizontal path have similar air masses, so the view here is through the equivalent of approximately 13 air masses.
Figure 2. The Miniature Integrated Speckle imaging Telescope (MIST) includes several co-aligned optical channels, cameras, video image processing, video recording, timing, laser control, targeting, and power management. It can be operated as a completely self-contained instrument or provide a rich set of support services to other external sensors and emitters.
Figure 3. This 60m (200 feet) tall radio tower on Catalina Island is at a distance of 77.4km (48.1 miles). The first 32km (20 miles) of the optical path is across the Los Angeles basin. This image was recorded with a 180mm aperture telescope near dusk with excellent seeing conditions. Six dish antennas, three beacon lights, and a pair of lightning rods are visible on the antenna mast. Also visible nearby are a pair of power poles, a building to the right of the antenna, and a fence surrounding the antenna base. The limiting (best) resolution in the scene is estimated to be approximately 0.5m (6μ rad).
Figure 4. This image of a 20m radome on San Clemente Island was recorded over a horizontal optical path of 128.3km (79.7 miles). The limiting (best) resolution in the scene is estimated to be approximately 2m (16μ rad).
Figure 5. This image of the third stage of a Minotaur space launch vehicle was acquired four seconds after third stage ignition. The viewing distance is 319km (172 nautical miles). Trailing by 140m (460 feet) is the spent second stage causing a faintly visible shock wave.
Step 1. Contrast enhancement: The contrast of distant scenery often is quite low because of aerosol scattering and attenuation. Most of the contrast enhancement is during the acquisition phase and includes the use of optical polarization, spectral long-pass filtering, and real-time analog video processing. Spectral long-pass filtering also reduces atmospheric turbulence blurring due to the r0 wavelength dependence.
Step 2. Isoplanatic search: The next step is to search for isoplanatic regions within the image sequence. A single-look speckle image sequence typically consists of 1000 images acquired at a rate of 60/s. The individual exposures are on the order of 1/100s or less, short enough to freeze the atmospheric turbulence induced speckle structures. Most of the images have significant image blurring due to the random overlapping atmospheric lensing of the turbulence cells. However, in a small percentage of the images, several small isoplanatic regions are occasionally much sharper than the average. Image sorting is used to select those images that include some high sharpness content in at least part of the scene. This is about 1% of the images during reasonably good seeing conditions. The percentage is lower for larger apertures and worse seeing. Although technically simple, this technique can be quite effective. Image sorting has been used to produce some of the sharpest ground-based planetary images made to date.4
Step 3. Isoplanatic merge: The third step is to merge the previously-identified isolated isoplanatic regions into a single composite image with increased total isoplanatic area. Essentially, the three-dimensional image cube is mapped onto a two dimensional surface by overlaying only the isoplanatic regions
Step 4. Image fusion: The fourth step is to combine separate images— often from different sensors, times, wavelengths, and polarization— into a single image with a higher information density.
MIST has proven to be a useful tool for atmospheric optics research, both as a standalone instrument and as a support device for other sensors. An image processing work flow has been established that at least partially compensates for atmospheric degradations and increases the information density. Additional hardware enhancements and software algorithm developments are planned for the future.