In order to solve the problem of fixed ambiguity and decreased accuracy in GNSS displacement monitoring of the offshore floating platforms, an attitude correction algorithm based on the fusion of a multi-antenna GNSS and an accelerometer was proposed using the Kalman filtering method. The algorithm was validated on a physical simulation platform and a real offshore floating platform. The results indicate that this fusion method effectively compensates for the loss of high-frequency displacement information caused by low GNSS sampling rates, improves situations in which the fusion effect deteriorates due to attitude changes, and enhances the accuracy of GNSS and accelerometer fusion monitoring through offshore buoy testing.