Based on a virtual structured-light data coding algorithm, three-dimensional (3D) data can be encoded into the phase of a two-dimensional (2D) color fringe image to complete the data compression. In the process of decoding a 2D color fringe image to 3D data, phase calculation and phase unwrapping are required. Due to the compression and storage of the 2D image, a quantization error exists at the edge of the phase index map, which will cause a local error in the unwrapping phase and will result in a 3D data decoding error in the z direction. To overcome this limitation, the decoding process of virtual structured-light 3D data is improved. The proposed algorithm can revise the local phase error and enhance the 3D data decoding accuracy efficiently. In the proposed algorithm, deviation between absolute phase unwrapping and relative phase unwrapping is calculated. Since the deviation is fixed, phase compensation is performed when the local deviation is changed sharply, and the local error of the decoded 3D data caused by the edge quantization error of the phase index map is suppressed. Experiments are conducted to verify the effectiveness of the proposed algorithm, which makes the virtual structured-light 3D data encoding algorithm more robust after compressing and storing the encoded 2D images. In comparison with the median filtering algorithm, the proposed algorithm can reduce the root-mean-square error of data decoding by an average of 9.7%.