This paper presents an error resilient coding scheme for atomic image bitstreams, as generated by Matching Pursuit encoders. A joint source and channel coding algorithm is proposed, that takes benefit of both the flexibility in the image representation, and the progressive nature of the bitstream, in order to finely adapt the channel rate to the relative importance of the bitstream components. An optimization problem is proposed, and a fast search algorithm determines the best rate allocation for given bit budget and loss process parameters. Simulation results show that the unequal error protection is quite efficient, even in very adverse conditions, and it clearly outperforms simple FEC schemes.