A new adaptive regression type of image destriping method is introduced to reconstruct failed lines in multispectral image data. The method uses available information from the failed pixel surrounding due to spectral and spatial correlation of multispectral data. The reconstruction is based on two competing adaptive regression models from which the locally optimal predictor is selected.

doi.org/10.1109/ICPR.1994.577144
12th IAPR International Conference on Pattern Recognition - Conference C: Signal Processing - Conference D: Parallel Computing, ICPR 1994
Centrum Wiskunde & Informatica, Amsterdam, The Netherlands

Haindl, M, & Šimberová, S. (S.). (1994). An adaptive image line reconstruction method. In Proceedings - International Conference on Pattern Recognition (pp. 153–155). doi:10.1109/ICPR.1994.577144