Embedded lossy image compression based on wavelet transform

This paper describes an algorithm for embedded lossy image compression based on the discrete wavelet transform (DWT) and the zerotree coding scheme (EZW). In this algorithm there are two lossy stages, a rounding operation after the transform and a truncation of the coding output. A comparative study of different filter banks (from different wavelet families and with different filter lengths) for the wavelet transform and different image scans for the encoding is presented. We show their impact in realistic image compression results for our algorithm, specially in compression ratio and reconstruction error.