A new trinocular stereo matching algorithm is proposed. The collinear trinocular stereo, which is an almost occlusion-free system, is adopted, and the classical binocular stereo algorithm, adaptive support weight, is used for window selection. A novel camera selection technique is presented to utilize the information of different target images more efficiently. Moreover, a trinocular disparity calibration algorithm is introduced, so the initial disparity can be refined and a precise result can be produced. Some image sequences of real scenes are used to evaluate our algorithm. The result shows that the proposed method can extract an accurate disparity map. It is concise and efficient.