Dead Zone Quantization in Wavelet Image Compression | Mini Project in Ece 253a |

A common quantizer implementation of wavelet image compression systems is scalar quantization followed by run-length and Hu man coding. The low complexity of such a quantizer system makes it suitable for hardware implementations. By replacing the scalar quantizer with a dead zone quantizer, the performance of the system in terms of PSNR/bpp increases. The purpose of this mini project is to investigate how big the dead zone should be for optimal performance and how much quality can be gained. It is found that for the test image used the optimal size is about 1.9 steps and that the quality increases about 0.5 dB.