Compression of color images using mixed transform techniques

A new technique for lossy compression of color images is presented. This technique is based on both the discrete wavelet transform and the discrete cosine transform. The wavelet coefficients are calculated using the cascade algorithm and coded using vector quantization and arithmetic coding. The coarsest scale wavelet coefficients containing the low frequency information are transformed using block DCT followed by scalar quantization. The proposed technique was tested using 256/spl times/256 aerial images. Compression ratios exceeding 60:1 have been obtained at a peak signal-to-noise ratio of above 30 dB without any perceivable degradation in image quality. The results obtained indicate that the proposed technique performs better than the JPEG standard.