Image compression using a mixed-transform technique and vector quantization

A new technique for lossy image compression is proposed. The technique is based on the wavelet transform of the original image combined with the discrete cosine transform of the low-frequency scaling coefficients. The remaining wavelet coefficients are compressed using vector quantization. The technique was tested on a series of facial images and the results obtained indicate an average peak-signal-to-noise ratio of 34 dB at an average compression ratio of 60:1. Comparisons with other known coders have shown that the compression performance of the proposed technique is better than that of the JPEG coder and comparable with that of the EZW and SPIHT coders.