Efficient VLSI Implementation of Embedded Zerotree Wavelet for Picture Transform

Wavelet Transform is a powerful mathematical tool for hierarchical multi-resolution analysis of functions/images and offers embedded lossy to lossless coding, progressive transmission by pixel accuracy and by resolution. To address these needs a number of recent embedded transform coders, including Shapiro´s Embedded Zerotree Wavelets (EZW) scheme employ a common algorithm called significance tree quantization (STQ) which are based on empirical work and a priori knowledge about transform coefficient behavior. Pure software implementations of the EZW, however appear to be the performance bottleneck in real-time systems. A design for efficient hardware acceleration of the EZW is prepared by utilizing stack based architecture and intelligent estimation of thresholds and is implemented at Register Transfer Level (RTL) in Verilog. The hardware can be used to accelerate multimedia applications such as JPEG2000 or MPEG-4. The hardware unit operates with a clock frequency of 250 MHz and showed a considerable speedup for a grey scale picture size of 512^*512 pixels.