IMAGE WAVELET TRANSFORMS IMPLEMENTED BY DISCRETE WAVELET CHIPS

Both discrete wavelet transform (DWT) and inverse DWT are implemented using the lossless quadrature mirror filter (QMF) bank. The image passing through the finite impulse response QMF filtering becomes blurred and thus requires fewer number of pixels. Such a decimation amounts to the critical sampling that leads to the complexity O (N) for N data. The data compression comes from the permissible bits per pixel dynamic range compression of those filtered images having fewer details. The image reconstruction at a telereceiving station is accomplished by means of the inverse DWT. Thus, a complete biorthogonal basis of QMF is implemented by a fast wavelet transform chip designed with Verilog HDL and the image processing is demonstrated numerically. Adaptive DWT is sketched.