Abstracto

A Memory-Efficient Stripe Based Architecture for 2D Discrete Wavelet Transform

M.Tharangini, V.Geetha and Dr.G.Murugesan

In this paper, a memory efficient new stripe based scanning method for 2- dimensional discrete wavelet transform is proposed in order to achieve an efficient memory and to reduce critical path. The efficient memory is achieved with a stripe based scanning method that enables tradeoff between external memory bandwidth and internal buffer size. Based on the data flow graph of the flipped lifting algorithm, the processing units are developed for maximally utilizing the inherent parallelism. With S number of processing units, the throughput can be scaled while keeping the latency constant. When compared with the existing architecture, the proposed architecture requires less memory. For an N × N image, the proposed architecture consumes only 3N + 24S words of transposition memory, temporal memory, and pipeline registers. The synthesized results in a Modelsim process show that it achieves better area-delay products than the best existing design where S is considered to be 2, 4, and 8.