User:TO ljk

=Co-registration of Optically Sensed Images and Correlation (COSI-Corr)=

Co-registration of Optically Sensed Images and Correlation, (abbreviated COSI-Corr), is a software package used in geodesy and remote sensing to generate maps of horizontal changes in the surface of the Earth (or even Mars). This software compares two or more optical images of a surface taken at different times, for example, before and after an earthquake. Although COSI-Corr was originally developed to measure co-seismic deformation, its application has expanded to measuring surface changes caused by volcanic eruptions, landslides, shifting sand dunes, ice flow velocity of moving glaciers, and even sand dune migration on Mars.

The input photographic images can be obtained either by satellites (e.g., SPOT and ASTER) or by aircraft. Unlike most previous techniques to measure surface change, COSI-Corr can compare an image obtained from a satellite with another obtained from aircraft, even if the photos were taken from different views. Moreover, it makes these comparisons using an automated procedure that takes just a few hours of processing time. COSI-Corr can potentially measure XXX-scale changes in the earth's surface with an uncertainty of just 1/10 pixel, over time-spans ranging from XXX to XXX.

Technique
Measuring ground deformation by comparing before and after photos can be very difficult. The two photos to be compared are usually taken from different camera positions and even with different cameras. There are parallax effects and distortions in the images due to the motion of the satellite carrying the camera. Thus correlating two images requires extremely accurate modeling of these distortions.

Previous techniques such as InSAR can reveal small changes in surfaces, but not large changes such as those caused by a large earthquake or fast-moving glaciers. Other techniques, such as sub-pixel correlation of pre- and post-earthquake optical images, suffer from numerous limitations, including uncertainties in the imaging systems (such as?) and the platform attitudes (what is this?), which bias the measurements.

COSI-Corr corrects for these limitations, while providing accurate estimation of sub-pixel displacement between images, through a new method of data processing. It provides quantitative surface dynamics measurements by automatic and precise ortho-rectification [define], co-registration [define], and sub-pixel correlation [define] of images, using an automated, four-step procedure. Three of these four steps involve correcting geometric distortions that arise when projecting a 2-D picture onto a 3-D surface. For these steps, COSI_Corr takes advantage of the availability of accurate digital elevation models with global coverage (Shuttle Radar Topography Mission).

Computer inputs are the raw images, and the orbital positions of the satellite and/or aircraft at the time of the photos, as well as the craft's orientation or tilt. The software then accurately superimposes the images (a process called ortho-rectfication). The result is photos that are sub-pixel correlated [define].

Step one in the procedure matches the coordinates of the ground with the coordinates of the photos. In other words, the images are re-projected onto a common geometry, called an "ortho-rectified" geometry, which takes into account the topography of the ground. While not a new task, COSI-Corr uses a new, more accurate method to do this. In step two, the software determines the position of the satellite in its orbit at the time it took the photo as well as its motion (is this something you have to input by hand, or does the software figure this out?). In the third step, the software correctly "drapes" the 2-D image onto the 3-D topography of the ground. And in the fourth step, it precisely combines the images (called co-registration) in order to accurately measure surface displacements.

Difficulties with COSI-Corr
•	The measurement depends on many factors such as clouds, snow, vegetation cover, shadows). •	Limitations arise when processing and analyzing a multitude of very large images and very-high-resolution satellite imagery (sub-meter resolution). •	Limited to optical images - plan to extend to radar images as well. •	Can be limited to availability of accurate digital elevation models, especially in mountainous areas

Advantages of COSI-Corr
•	Can use any two images •	No limit on how large the displacement is •	An enormous amount of satellite data exists in archives and continues to be acquired.

Applications
•	Measure ground deformation caused by earthquakes •	Measure ice flow velocity in moving glaciers •	Measure speed of sand dune migration on Mars •	Measure slowly moving landslides •	Volcanoes