An illumination method based on binary spatio-temporal encoding for 3D imaging is proposed. Every row of the projection plane is divided into some sections. The code of a section is encoded with the spatial and temporal coordinates of pixels in the section. During measurement, the codes of sections are retrieved by analyzing the captured image sequence. The adjacent codes with given number are assembled to form subsequences after all the codes of sections are recovered. The corresponding points between surface of scene, projection plane and camera image plane are worked out by matching subsequence in the designed sequence. Then, the shapes of objects are reconstructed with triangulation. The experimental results reveal that dense depth images can be obtained and texture can be acquired simultaneously without any additional images. The proposed method is robust and accurate.