Coral reefs are complex marine ecosystems and environment-sensitive that are constructed and maintained by biological communities that thrive in the warm tropical ocean. Remote sensing provides a more efficient way to observe and monitor shallow coral reefs than alternative field survey. Dong-Sha Atoll is located at the northern continental margin of the South China Sea. The area of the atoll is more than 300 km2 with an average depth of 10 m, and has being abused by destructive fishing during recent decades. In this paper, three satellite broadband multi-spectral imageries (Quickbird2, ETM+, and SPOT5) are used to survey this area. In the preprocessing, low-resolution images are geometrically corrected to high-resolution imagery. For information extraction, unsupervised classification functions are adopted for ETM+ and SPOT5 data, and supervised classification method for Quickbird2. The classification results are (or not) merged into coral reef, and operated by vectorization, simplification, and topological analysis. There are 1331 coral reefs larger than 100 m2 with the detection limitation of 3×3 pixels at multi-band data of Quickbird2. The results extracted from SPOT5 data are better than that of ETM data at silty lagoon due to its higher resolution, and less at atoll reef bodies in the absence of blue waveband.