Multichannel Restoration Techniques Filip Sroubek Many degradation phenomena that occur in imaging systems are modeled by a convolution with a blur mask. Many singlechannel restoration (deconvolution) techniques were proposed with satisfying results in the past. However, the singlechannel blind deconvolution (blur masks are unknown) is feasible only if several strong constraints are imposed. Treating the deconvolution problem in the multichannel framework exhibits very promising features. In general the multichannel problem is better conditioned and thus the estimated original image is less effected by noise. An overview of multichannel restoration techniques will be given and single-multichannel relations will be mentioned. Intrinsically multichannel methods will be emphasized and above all the subspace technique based on the greatest common divisor will be discussed in detail.