An image discrete wavelet transform and the hardware implementation
Charles Chia-chuen Hsu, Jinghua Ding, Mona Elwakkad Zaghloul, Harold H Szu · 2002
Both the Discrete Wavelet Transform (DWT) and inverse DWT are implemented by using a lossless Quadrature Mirror Filter (QMF) bank which approximates a Gaussian function. The image passing through the finite impulse response (FIR) QMF filtering becomes blurred and thus requires less number of pixels. Data compression comes from the permissible bits per pixel (bpp) dynamic range and the compression of those filtered images having lesser details. The image reconstruction at a telereceiving station is accomplished by means of the inverse DWT. A Mexican-hat like function is used as a filter in the DWT. The numerical simulation is given and verified. A hardware implementation of the Mexican hat function is included. The SPICE simulation of the CMOS circuit to implement the Mexican hat function is also demonstrated.