Image enhancement can be one of the most important tools in the processing of large data sets, making the results more suitable for classification than the original data are as, for example, two-dimensional CCD images of galaxies. One possible approach for image enhancement can be subdivided into two categories: the spatial processing, in which the pixels in the image are manipulated directly, and the transform on the spatial frequency domain, where the image function is first transformed to the transform domain and then processed to meet a specific problem or behavior requirements. After this processing, the inverse transform is needed to yield the final spatial image results.
From an intuitive point of view, an image comprises both high and low spatial frequencies, where the high spatial frequencies correspond to sharp edges and the low spatial frequencies correspond to those regions of approximately uniform gray level.
Pioneering works on studying the observed structure of some galaxies have already been done, even in the mid-century as, for example, Danver's (1942) study of the mathematical form of the spiral arms in galaxies, the Fourier analysis of spiral observed structures by Kalnajs (1974), and Aoki et al. (1979), who predicted that multiarmed modes should grow in some galaxies in addition to two-armed spirals, and suggested that the coexistence of several spiral waves is possible. Spectral analysis of spiral and bar structures in galaxies were developed through two-dimensional Fourier analysis by Iye et al. (1982) and Iye (1983). Fourier transforms on HII distribution and also on spiral component in galaxies have also been performed by Considère & Athanassoula (1982 and 1988, respectively, and references quoted therein; see also Considère 1980).
Likewise, filtering is one of the most common processes used in the transform domain for image enhancement. The filtering method is applicable to problems such as the recognition of finger prints and visual filtering of letter identification as Solomon & Pelli's (1994) and Secker's (1995) works, the latter having applied ring median filter on digital images (see also Harris et al. 1983). Sulentic et al. (1985) have pointed out the advantages and disadvantages of spatial frequency filtering in comparison to nonlinear domain tools, e.g., median, modal, and mean filters. The number of textbooks on image processing and restoration is not depreciable; the reader is referred to Marion (1991) and Bates & McDonnell (1989).
In this work the idea of spatial filtering to find fine structures in galaxies is tested. The computing employs some packages of IRAF and STSDAS and those operations are quickly and easily done. The filters to create the transfer function are simple forms of high- and low-pass filters. The creation of a Fourier hologram image to be used as transfer function after simple filtering is briefly discussed.