A simple three-dimensional (3D) reconstruction algorithm is designed based on the rule of horopter and the theory of binocular stereo vision. The optical axes of the two cameras to shoot the same object are cross-placed. The image pixel coordinates are changed into spherical coordinates. The feature points are matched in spherical coordinates domain and the precise points are gotten by curve fitting. Finally, we turn the points to space coordinates, and obtain the 3D model of the target object. Experimental results show that this matching algorithm can accurately determine the location of the target object, and it has low computing complexity. The data processing of the later stage adopts the method of piecewise fitting so that the points on the 3D model of the target object obtained after the coordinate transformation distribute accurately and densely.