Automatic Registration of Optical Images, a Stake for Future Missions: Application to Ortho-Rectification, Time Series and Mosaic Products

Today, new earth observation missions are designed from satellite to ground segment to best fit the end-user needs while reducing the overall costs and complexity. Furthermore, an efficient use of high resolution image data is only possible if its location is accurate enough. The accuracy of the image location model depends on the knowledge of platform and orbital parameters (instruments calibration, satellite orbit and pointing) and also of the ground elevation (Digital Elevation Model). Good accuracy can be ensured by stringent requirements on onboard equipments (GPS, gyro, ...) but also by appropriate ground processing which is easier and less expensive to setup. This paper describes and evaluates an automatic method to improve location models during ground processing. This method has been implemented and tested at CNES for current and new missions (such as SPOT/FORMOSAT image ortho-rectification, Pleiades-HR images mosaicking or Venmus images time-series generation). These results must be considered for future missions designing.