Three novel image compression schemes are proposed here for multimedia applications. The schemes are designed to combine the advantages of vector quantization and block truncation coding based on quadtree segmentation. A new vector quantization scheme, called subsampling VQ (SVQ), is proposed and used in the new schemes. The spatial correlation between neighboring image blocks and uniformity in variable-sized image blocks is exploited in these schemes. Simulation results show that the new schemes have succeeded in achieving better coding performance with subjective image quality at low bit rates.