LMMIP

IMAGE RECONSTRUCTION TECHNIQUES BASED ON PATTERN RECOGNITION THEORY
Main idea: reconstruction by " precedent": a reconstruction algorithm developed on the basis of a reference pair "ideal image - distorted image"
is then applied to other same-class images

Schematic view of data transformation
a) learning

      1. Distortion system model;
      2. Feature generation;
      3. Classifier;
      4. Error estimation.
    b) processing

        1. Feature generation;
        2. Classifier;
Results of processing:
Original image
Distorted image
Reconstruction
by Wiener filter
Reconstruction by use of pattern
recognition technique

Publications:
1. V. V. Sergeyev, A. V. Chernov, Pattern Recognition and Image Analysis. 1997, v.6., N.4, pp. 474-479.
2. V. V. Sergeyev, A. V. Chernov, Pattern recognition and image analysis: novel information technologies, 3-d all-Russian and CIS conference, Nizhny Novgorod, 1997, Conference Proceedings in two parts; Part II, pp. 3-4.

LMMIP