小波与傅里叶分析基础第12讲.ppt

12 Key words and expressions initialization iteration termination downsample upsample high-pass filters low-pass filters 5.2 Implementing Decomposition and Reconstruction 5.2.1. The Decomposition Algorithm Initialization First, we have to decide on the approximation space Vj that best fits the information available on f. Second, we need to choose to best fit f itself. Theorem 5.6 Let be a given multiresolution analysis with a scaling function that is compactly supported. If is continuous, then, for j sufficiently large, where . Iteration Let h and l be the sequences Definition 5.2 If then its downsample is the sequence or for all . Termination The two filters that we use, h and l, are called the decomposition high-pass and decomposition low-pass filters, respectively. Convolution form: Operator form: Example 5.6 5.2.2 The Reconstruction Algorithm Iteration Let and be the sequences Termination Definition 5.3 Let be a sequence. We define the upsampling operator U via or The reconstruction algorithm can now be simply formulated in terms of discrete filters. Convolution form: Operator form: We remark that and are called the reconstruction high-pass and low-pass filters, respectively. 5.2.3 Processing a Signal Sample. Decompose. Process the signal. Reconstruct. Example 5.7 The End * * Initialization *