Novel architectures for the lifting-based discrete wavelet transform

In this paper, we propose two modular and highly efficient architectures, the recursive architecture and the double scan architecture, for implementing the lifting-based discrete wavelet transform. The recursive architecture simultaneously processes all stages of the transform, while the double scan architecture processes two independent signals at the same time. Compared to other known architectures, both new architectures significantly improve the efficiency and hardware utilization. Compared to other published lifting-based architectures, the new architectures speed up the computational time by up to 100% with either simpler or comparable hardware complexity.