Summary form only given. The fast Hartley transform (FHT) is used for the progressive reconstruction of an image in a still image retrieval system with a narrow bandwidth channel. Due to its real-valued nature, FHT has the advantage over the fast Fourier transform (FFT) in the computation speed, and it has the identical forward and the inverse transform kernel, among other properties, so it is a good substitute for FFT in many applications. However, for the 2D transform, the FHT is a little awkward compared with the FFT. For the 2D FFT, it is possible to split the transform into two 1D FFTs. The Hartley transform, whose kernel function is sine plus cosine, cannot be split in the same way as the FFT. The authors have circumvented this limitation by using multiple 1D FHTs to perform a 2D transform. They call this transform the pseudo-2D Hartley transform.
展开▼
