The GF-7 satellite is the first sub-meter two-line-array stereo imaging satellite of China. It is equipped with two sets of laser altimeters and laser footprint cameras to capture multi-source remote sensing data simultaneously. In this paper, the multi-source remote sensing data of GF-7 satellite were used to promote the horizontal and vertical accuracy, which used laser altimetry data to optimize the vertical accuracy by skewness, median, linear polynomial and quadratic polynomial model, and the footprint image was used to optimize the horizontal accuracy in the first-order affine transformation method. Moreover, the accuracy of uncontrolled plane-elevation, laser elevation optimization, footprint-laser plane-elevation optimization and field-laser plane-elevation optimization were evaluated through the field control points. The experimental results show that the vertical accuracy of DSM can be improved significantly with the support of laser altimetry data. The mean vertical error of the DSM produced without GCPs is -4.268 m and the mean square error is 4.518 m. While the mean vertical error and mean square error of the DSM optimized by the median model are improved to -0.272 m and 1.508 m, and the mean vertical error and mean square error of the DSM optimized by linear model reach -0.320 m and 1.351 m. The horizontal accuracy of DOM can be improved by the footprint image. The mean horizontal error is optimized from 13.606 m to 5.341 m, and the mean square error is optimized from 13.626 m to 5.495 m.