POINT-SOURCE LOCALIZATION IN BLURRED IMAGES BY A FREQUENCY-DOMAIN EIGENVECTOR-BASED METHOD

In this paper, we address the problem of resolving and localizing blur red point sources in intensity images. Telescopic star-field images bl urred by atmospheric turbulence or optical aberrations are typical exa mples of this class of images. A new approach to image restoration is introduced, which is a generalization of 2-D sensor array processing t echniques originating from the field of direction of arrival estimatio n (DOA). It is shown that in the frequency domain, blurred point sourc e images can be modeled with a structure analogous to the response of linear sensor arrays to coherent signal sources. Thus, the problem may be cast into the form of DOA estimation, and eigenvector based subspa ce decomposition algorithms, such as MUSIC, may be adapted to search f or these point sources. For deterministic point images the signal subs pace is degenerate, with rank one, so rank enhancement techniques are required before MUSIC or related algorithms may be used. The presence of blur prohibits use of existing rank enhancement methods. A generali zed array smoothing method is introduced for rank enhancement in the p resence of blur, and to regularize the ill posed nature of the image r estoration. The new algorithm achieves inter pixel superresolution and is computationally efficient. Examples of star image deblurring using the algorithm are presented.