Abstract: In order to improve defects of the measurement system based on phase measuring profilometry, such as low measuring efficiency and bad system stability etc, a fast phase-based stereo matching method based on improved LMedS algorithm and greedy strategy was presented.The adjustment models was introduced into the LMedS algorithm to search the optimal model parameters.The accumulation of computing times of the cost function was cut down by eliminating the false samples as soon as possible, which kept the better robustness of LMedS and reduced its computation amount effectively.Then, the initial inliers obtained by the improved LMedS were optimized with the greedy strategy.Those relatively inappropriate inliers were removed and the optimized inliers were used to compute the fundamental matrix to enhance the precision and improve the robustness of the algorithm.Finally, through three-dimensional reconstruction of certain objects, this paper compared the performances of the traditional phase-based stereo matching method based on camera calibration result and the improved method.Experimental results showed that the proposed algorithm can met the requirements for real-time, precision and robustness in the fields such as three-dimensional measurement and the data was complete and reliable.