Research on target measurement based on fusion of binocular and TOF system

Traditional target detection mostly uses manual detection, which has poor universality and takes a long time. It is difficult to measure some large-scale three-dimensional objects. Therefore, the use of three-dimensional vision for object detection has become a hot spot. According to the complementary advantages of binocular system and TOF system, the two systems are combined effectively. The coordinate relation between TOF system and binocular system is established by joint calibration. The data obtained by the TOF camera were mapped to the camera’s perspective on the left side of the binocular system to obtain the TOF parallax map under the camera’s perspective on the left side of the binocular system. The parallax map was transformed into the resolution of binocular image through image processing. The concept of confidence degree is put forward. According to the confidence degree, the different system weights of data fusion are determined. After fusion, the matching accuracy is improved, and fast 3D reconstruction data acquisition can be completed.