Naslov
Single Frame Multichannel Blind Deconvolution by Non-negative Matrix Factorization with Sparseness Constraint

Autori
Kopriva, Ivica

Izvornik
Optics Letters (0146-9592) 30 (2005), 23; 3135-3137

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
Signal Processing; Deconvolution; Image reconstruction-restoration; Inverse problems

Sažetak
Single frame multichannel blind deconvolution is formulated by applying bank of Gabor filters to blurred image. The key observation is that spatially oriented Gabor filters produce sparse images and that multichannel version of the observed image can be represented as a product between unknown non-negative sparse mixing vector and unknown non-negative source image. Therefore blind deconvolution problem is formulated as non-negative matrix factorization problem with sparseness constraint. No a priori knowledge about the blurring kernel and original image is required. Our good experimental results demonstrate viability of the proposed concept.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika

POVEZANOST RADA