In this paper we present an adaptive image coding algorithm based on novel backward-adaptive quantization/classification techniques. We use a simple scalar quantizer to quantize the image subbands. Our algorithm uses several contexts to characterize the subband data and different arithmetic coder parameters are matched to each context. We show how the context selection can be driven by rate-distortion criteria and how the performance can be improved by replacing the scalar quantization strategy by a entropy-constrained approach. Our results are comparable or better than the recent state-of-the-art with our algorithm also having advantages in terms of simplicity.