====== Image Restoration ======

The image restoration problem attempts to recover image that has been degraded by the limited resolution of the sensor as well as by the presence of noise. The resolution of images obtained by the satellite sensors is degraded by sources such as: optical diffraction, detector size and electronic filtering. As a consequence, the effective resolution is, in general, worse than the nominal resolution, that corresponds to the detector projection on the ground and does not take into consideration the sensor imperfections. Through restoration techniques, it is possible to improve image resolution up to a certain level.

The restoration method studied here is based on the Modified Inverse Filtering (MIF) and explores the idea of combining the restoration process with an interpolation process to generate images with a better resolution over a finer grid.
The design of the MIF filter requires knowledge of the transfer function of the sensors which is approximated as a separable Gaussian function. 

===== Restoration Examples =====

  * [[leila:research:image_restoration:cbers_ribeirao|CBERS 2 - CCD Ribeirão Preto]]
  * [[leila:research:image_restoration:landsat5|Landsat 5]]
  * [[leila:research:image_restoration:landsat7|Landsat 7]]

===== Software =====

The main interface of the system is shown at the bottom. The user can choose between the CBERS-1, CBERS-2, SPOT, TM-5, ETM+ and MODIS sensors in the Sensor option. Then, one uses the Browse... button to select one file (each file must consist of a single spectral band). Then the user chooses the output resolution using the Pixel menu. This menu can be used to generate images with a different output pixel size than the input value (for example, to produce an interpolated 10 m CCD band from an original 20 m image). The Band menu allows a choice between the different spectral bands of the image. The EIFOV menu allows the choice of the estimated EIFOVs in the directions X and Y. Their values are changed just to test various hypotheses concerning differences between actual and estimated EIFOV.

  * [[http://www.dpi.inpe.br/~leila/downloads/restau_05052009.exe|Download the system]]

{{leila:research:image_restoration_example.png|}}