A Flexible Floating-Point Wavelet Processor

In this paper, we propose a floating-point lifting-based wavelet processor that can perform various forward and inverse DWTs. Our architecture is based on processing elements that can perform either prediction or update on a continuous data stream in every two clock cycles. As wavelet transforms are not only used in JPEG2000, their wavelet filters and the corresponding transformations cannot be satisfied by using integer arithmetics.For this purpose, IEEE 754-compliant floating-point arithmetics are used to compute the transformation.We feature a multi-context configuration to select among various DWTs. Different memory sizes and multi-level transformations are supported. For the 32-bit implementation, SNR values between 124-144 dB can be achieved. The estimated area of the proposed wavelet processor with 2×2×512 words memory and 8 PEs configuration in a 0.18-¿m technology is 8.3 mm square and the estimated operating frequency is 351 MHz.