This paper presents an unequal error protection scheme for atomic image bitstreams. An atomic stream is the encoded version of a digital image, which is represented as a sum of bi-dimensional functions, as typically generated by Matching Pursuit encoders. The atomic structure of the compressed image presents an enormous advantage in terms of flexibility, since any atom of the stream can receive a different treatment, like a finely adapted protection against error. We take benefit from this property to propose a joint source and channel coding algorithm, that finely adapts the channel rate to the relative importance of the bitstream components. A fast search algorithm determines the distortion-optimal rate allocation for given bit budget and channel loss parameters. We further extend the algorithm to differentiated protection of region of interests. Simulation results show that the unequal error protection is quite efficient, even in very adverse conditions, and it clearly outperforms simple FEC schemes.