Multi-view vision technology is use of multiple images to reconstruct three-dimensional (3D) information of the research object, and the images were captured with more than two cameras from different angles. This technology has the advantages of high efficiency, simple structure and low cast. It is very suitable for on line noncontact product test and quality control. The existent multi-view vision technology is focus on the visible information, thus it is easy to be influenced by the testing environment (weather, background, light etc.), so it’s application field has some limitations. However, in the field of medical diagnostic technology, infrared vision technology reflects its advantages, like determining whether the tissue lesions by observe certain parts temperature distribution of the body. This paper is studied on infrared Multi-vision which is based on visible binocular vision theory. Firstly, obtained the intrinsic parameter and external parameter of each infrared thermal imager by Zhang’s calibration method; Secondly, taken infrared images from different angles, then combined the infrared images two by two to do feature point detecting and matching in order to find the points to be reconstructed; Finally, reconstructing 3D profile information based on calculating point clouds of the spatial coordinates.