A novel image compression scheme based on absolute moment block truncation coding (AMBTC) is proposed. The quadtree segmentation technique is employed in this scheme to exploit the variable-block-sized segmentation. In addition, the concept of using visual patterns to reduce the storage cost of the BTC bit plane is incorporated into this scheme. Finally, the bit plane omission technique is employed if these two quantization levels in each image block are quite similar. According to the experimental results, it is shown that the proposed scheme requires very little computational complexity. Besides, a good image quality of the compressed image can be obtained while the required bit rate is relatively low. In other words, the proposed scheme indeed provides a good approach to compress digital images with little computational cost at a low bit rate.