Objective  The triple scan of the binocular structured light sensors is based on the binocular vision reconstruction of fringe projection, with additional point cloud reconstructed from the monocular vision system consisting of the left or right camera and the projector. Therefore, it has the advantage of better point cloud integrity in the measurement applications of reflective and uneven objects. However, because of the change of ambient temperature which exerts temperature drift impact on the projector, the binocular reconstruction point cloud and monocular reconstruction point cloud are stratified, resulting in the wrong point cloud reconstruction results of the triple scan. To solve this, a temperature drift online compensation method based on the orthogonal fringe projection is provided for the triple scan system of binocular surface structured light.

  Methods  A temperature drift compensation method based on the orthogonal fringe projection and binocular points constraint is provided in this paper. By projecting the orthogonal fringe patterns, the pixel coordinate in the projector image of points which are in the binocular and monocular imaging area can be determined simultaneously. Therefore, the accurate mapping relationship of binocular points in the image of the projector is established (Fig.3). Then based on the objective function of minimizing the reprojection error of the binocular reconstruction point in the projector image, the optimal external parameters of the projector after temperature drift compensation are solved to compensate temperature drift of the projector.

  Results and Discussions   In order to demonstrate the effectiveness and practicability of the proposed method, metal spheres and car threshold base parts were tested. In metal spheres experiments, the spherical center distance between binocular point cloud and monocular point cloud at different temperature is used to quantitatively evaluate the temperature drift level (Tab.1-2). When the temperature drift of the sensor occurs at 28 ℃, the values of spherical center distance on average are 1.89 mm, 1.48 mm, 1.73 mm, 1.67 mm and drop to 0.36 mm, 0.37 mm, 0.36 mm, 0.37 mm after the method is proposed. In car threshold base parts experiments, the Euclidean distance between binocular points and monocular points corresponding to the same pixel is shown (Fig.8). Both results show that temperature drift is decreased after the method is proposed with the temperature drift value reduced by 78.2% and 94.3% on average.

  Conclusions  Focusing on the issue that binocular and monocular point clouds are stratified in the application of the triple scan due to the change of the ambient temperature, the temperature drift online compensation method for triple scan measuring system of binocular surface structured light is proposed. Orthogonal fringe patterns are projected and meanwhile are captured by the binocular cameras. The accurate pixel coordinate in projector image coordinate system of binocular points are acquired by calculating the phase coordinate of every pixel in the camera image and projector image. According to the the constraint of the mapping relationship between binocular points and projector pixels, the external parameters of projector are calculated via minimizing the reprojection error of the binocular points in the projector image to compensate the temperature drift of the projector. The experimental results show that the proposed method can significantly decrease the temperature drift with the value reduced by 78.2% and 94.3% on average, which verifies the effectiveness of the method. This method can provides guarantee for the stability of triple scan and make the application of triple scan a reality.