In this work, near-lossless compression yielding strictly bounded reconstru
ction error is proposed for high-quality compression of remote sensing imag
es. A classified causal DPCM scheme is presented for optical data, either m
ulti/hyperspectral three-dimensional (3-D) or panchromatic two-dimensional
(2-D) observations. It is based on a classified linear-regression predictio
n, followed by context-based arithmetic coding of the outcome prediction er
rors and provides excellent performances, both for reversible and for irrev
ersible (near-lossless) compression. Coding times are affordable thanks to
fast convergence of training. Decoding is always real time. If the reconstr
uction errors fall within the boundaries of the noise distributions, the de
coded images will be virtually lossless even though encoding was not strict
ly reversible.