Compression of rotation images based on the wavelet transform
Kai Yang, Sentarou Okamoto, Hiroyuki Kudo, Takahiro Saito · 2002
A scheme for compression of rotation images is proposed. First, we take a wavelet transform to decompose an image using a biorthogonal filter bank. Second, according to Shannon's rate distortion theory, an interband vector quantisation (VQ) is developed, with the exception that if gain is below a threshold, it is simply considered as a zero vector. Third, Huffman encoding is applied to vectors, and scalar quantisation is applied to the low passed signal at the coarsest level. Finally, rate-distortion performance tests for rotation images and Lena image are performed.