ABSTRACT
Spectral analysis tools like Fourier transform and wavelet transform have been widely used in many engineering fields. One such example is image processing, where images are taken as real functions on a two-dimensional (2D) domain. Spectral analysis is to analyze and process functions in frequency domains, or in another saying, at different scales. It brought fundamental applications to image processing, such as feature detection1, smoothing, denoising, and compression. These low-level applications solidly support the high-level problems of the frontier research: object recognition, image retrieval, and clustering, to name a few.
