一种新颖实用的基于视觉导航的三维重建算法根据立体视觉在视觉导航应用中的特点，提出了由视差图重建三维场景的整套算法。首先，推导视差图里各点在摄像机坐标系下的三维坐标。其次，建立摄像机坐标系与车体坐标系的转换关系，得到更有实际意义的在车体坐标系下各点的三维坐标。最后，通过分格显示地貌的算法，大大减少了数据量。实验结果表明，该算法计算量小、鲁棒性强，完全满足视觉导航中的实时性、准确性要求。From the specialties of the stereo vision applied in vision— based navigation，we propose a whole solution to reconstruct 3D scene from the disparity map．First，we discussed how to get the camera— bas ed coordinates form the disparity map．Second，we developed the conversion from the camera— based to the vehicle— based co rdinates，thus，more practical reconstruction result was obtained．Finally，the quantity of data Was reduced greatly by subdividing whole scene into smaller subunits．The experiment results prove that our algorithm con meet the vision— based naviga—tion’S request for real—time an d validity ．