Details

Autor(en) / Beteiligte

• Park, Kwanjun
• Yang, Taeseok Daniel
• Kim, Hyung-Jin
• Kong, Taedong
• Lee, Jung Min
• Choi, Hyuk Soon
• Chun, Hoon Jai
• Kim, Beop-Min
• Choi, Youngwoon

Titel

Inversion-free image recovery from strong aberration using a minimally sampled transmission matrix

Ist Teil von

• Scientific reports, 2019-02, Vol.9 (1), p.1206-1206, Article 1206

Ort / Verlag

England: Nature Publishing Group

Erscheinungsjahr

2019

Quelle

Elektronische Zeitschriftenbibliothek - Frei zugängliche E-Journals

Beschreibungen/Notizen

• A transmission matrix (TM), a characteristic response for an input-output relation of an optical system, has been used for achieving diffraction-limited and aberration-free images through highly-aberrant imaging systems. However, its requirement of acquiring a huge-size TM along with its heavy computational load limit its widespread applications. Here we propose a method for TM-based image reconstruction, which is more efficient in terms of data manipulation and computational time. Only 10% of the TM elements for a fish-eye (FE) lens with strong aberration were sampled compared to that required for the image reconstruction by the conventional inversion method. The missing information was filled in by an iterative interpolation algorithm working in k-space. In addition, as a replacement of the time-consuming matrix inversion process, a phase pattern was created from the minimally sampled TM in order to compensate for the angle-dependent phase retardation caused by the FE lens. The focal distortion could be corrected by applying the phase correction pattern to the angular spectrums of the measured object images. The remaining spatial distortion could also be determined through the geometrical transformation also determined by the minimally sampled TM elements. Through the use of these procedures, the object image can be reconstructed 55 times faster than through the use of the usual inversion method using the full-sized TM, without compromising the reconstruction performances.