| Abstract: | The abrupt boundary truncation of an image may cause bright strips on a 2-D power spectrum in frequency domain and distortion of patterns in the frequency domain. The traditional solutions, including zero-padding and extension of the image by mirroring the data inside image, are useable for processing relatively simple, regular and stationary images. These methods, however, are not effective for dealing with images with holes or highly irregular shapes. This paper proposes an algorithm of using decay functions to handle boundary effects especially when filtering operation is involved in image processing. To validate the algorithm, we first employ a geochemical image interpolated by data set from a mineral district in Nova Scotia, Canada to analyze the effects of decay function on boundary truncations with the change of extension width. We then apply the algorithm to a real-life remotely sensed image. It is demonstrated that decay functions give superior overall performances over the traditional method.
|