In the past decades, digital image processing technologies have been widely applied in scientific research, industrial and agricultural production, as well as medical fields. And the research on image restoration technology is very important in the field of digital image processing.

In order to address the challenge of degraded image quality caused by the aberration of imaging system, the motion dithering and the atmospheric disturbance，a research group of Optoelectronic Detection and Signal Processing from Institute of Optics and Electronics（IOE）, CAS made great progress using improved multiscale, variational reasoning, parameter adaptive image restoration technology. The group achieved a high quality restoration of the blurred images by a series of models techniques, includinge degradation processes model, system imaging model, large-scale optimization calculation of ill-posed inverse problems, high performance real-time processing techniques, etc. Particularly, the variational estimation and adaptive modeling based on the spectral index is the most important, which can effectively overcome the ill conditioning of restoration of the inverse problem solution.

The researchers also conducted the research work on real-time implementation technology based on the GPCPU processing platform, and achieved near diffraction limited restored images with the rate of 50 frames per second in real-time..

The results were published in international journals and conferences, such as OSA conference and SPIE conference.

Raw image      Degraded image    RLA（Maui Air Force Base）

ISRA（New York University）   MIA（IOE 2005）    SBBD（IOE 2016）

Power spectrum: frequency component distribution