1

Topic: Introduction

Welcome to this COSI-Corr message board/discussion group. COSI-Corr stands for Co-registration of Optically Sensed Images and Correlation. It is a software package developed in IDL (Interactive Data Language), and integrated under the user friendly interface ENVI (Environment for Visualizing Images), by ITT.
COSI-Corr is provided free of charges for academic and research purposes, and can be downloaded from this address:
http://www.tectonics.caltech.edu/slip_h … tware.html

COSI-Corr provides tools to accurately orthorectify and co-register pairs of optically sensed images, irrespective of the sensor type and resolution. As long as a sensor is supported in COSI-Corr, all images between the supported sensors can be accurately orthorectified and co-registered. It is for example possible to co-register series of SPOT images, or series of aerial photographs, and it is also possible to register series of aerial photographs with series of SPOT images, etc...
COSI-Corr also provides a powerful correlation tool, which allows, for instance, for the measurement of ground deformation. The typical uncertainty on the displacement measurement is on the order of 1/10 of the nominal image pixel size. The measurement accuracy obviously depends on many factors (cloud, snow, and vegetation cover, shadows, temporal changes in general, steadiness of the imaging platform, defects of the imaging system, etc...), but in practice, 1/10 of the pixel size has been found to be an overall reasonable answer to "what can I expect from these images".
Although COSI-Corr was initially developed with the idea of measuring coseismic deformations, it has evolved to a more general tool. Typical uses are encountered in tectonics, of course, but also in glaciology to measure glacier flow velocities. Studies have also reported successful measurement of landslides motion, and sand dune migration rates. The quality of the co-registration is also welcome when studying land cover changes, for instance.


COSI-Corr’s design started in 2003 with the PhD thesis of Sebastien Leprince, and shortly after, its operational development was started by engineer scientist Francois Ayoub, under the supervision of Prof Jean-Philippe Avouac, at the California Institute of Technology. This project has been made possible thanks to NSF grants (EAR 0409652 & EAR 0636097), and thanks to funding from the Gordon and Betty Moore Foundation via the Caltech Tectonics Observatory.

Several people have contributed to the development of COSI-Corr. As of today, the principal contributors have been Sebastien Leprince, Francois Ayoub, and Jean-Philippe Avouac. Recently, Lionel Keene joined us as a full time scientific developer, increasing our work force. Other people who have been on the team for limited periods have also influenced our work, such as Sylvain Barbot and Pablo Muse. Extended discussions with Remi Michel and Renaud Binet have also significantly impacted our views. We will try to maintain an updated list of contributing people here:
http://www.tectonics.caltech.edu/slip_h … eople.html

In an effort to keep the community of interested users aware of "what's going on under the hood", we're trying to publish the technical details of the algorithms implemented in COSI-Corr. An up-to-date list of publications is available here:
http://www.tectonics.caltech.edu/slip_h … ences.html


By starting this forum, we expect to reach several goals:

1- We receive many similar questions from users. We hope to make available to all the answers to the most common questions. Simply put, this forum might be a live backup to the COSI-Corr user's manual, which many seem to find too brief.

2- As the community of users is growing, potential applications are growing as well. We therefore expect more direct feedback from the users, so that future releases can be made with less biased (ours) views.

3- We hope to provide a discussion area about remote sensing in general, regarding both technical problems and applications. Technical problems typically arise from a practical need for a specific application, and in the design of COSI-Corr, we have found that the technical and application communities were unfortunately often quite disjoint. We hope this forum will become a space for mixing ideas, reaching between applications and techniques.