| The three-dimensional reconstruction technology has a wide range of application requirements in national defense,aerospace exploration,industrial inspection,virtual reality and other fields.Binocular stereo vision technology is an important means to restore the three-dimensional information of an object through two-dimensional images.The binocular stereo vision technology of visible band is well developed,but this type of system is dependent on the environment,especially has a high requirement for illumination conditions,and could not be applied under smoke,heat and other special circumstances.Long wave infrared radiation reflects the temperature information of the object itself,in the dark environment can clear imaging,and can penetrate the smoke barrier,which provides a new way to expand the application field of binocular stereo vision system ? Based on the previous work,a three dimensional reconstruction technique was put forward which relies on the dual band point cloud integration of infrared and visible band in three dimensional space,and explores the main technical details.The contents of the study include the joint calibration,stereo matching and point cloud extraction,point cloud preprocessing,point cloud registration,point cloud integration and point cloud triangulation.The main research contents are as follows:1.Proposed a joint calibration method of infrared and visible light cameras.According to the characteristics of “ bi-binocular” stereo vision system and the requirement of the following works for the internal and external parameters of the camera set,a calibration target was developed which can be used in infrared and visible light band simultaneously based on the Zhang Zhengyou camera calibration method.Using the Maximally Stable Extremal Regions algorithm,the calibration points was successfully extracted from the calibration image in both visible band and infrared band,obtained the internal and external parameters of the cameras,proved that this method can achieve the calibration accuracy of the traditional grid target in the visible band and improved the calibration accuracy in the infrared band.2.Realized the stereo matching of the dual band image,and got the corresponding point cloud.Through the semi global stereo matching algorithm,the stereo matching of infrared and visible binocular images was completed,and obtained the corresponding disparity map.Through the Snake algorithm the segmentation of the disparity map was completed.Based on the segmented disparity map,two sets of point clouds corresponding to the infrared image and visible image are generated.3.Realized the registration and integration of dual band point clouds.The initial registration of the two groups of point cloud was completed by the calibration parameter of the left infrared and visible cameras.The precise registration of point cloud was accomplished through ICP algorithm.By means of the Milti Point Cloud Integration Algorithm Based on Normal Constraint,the integration of both clouds was completed,effectively reduced the Stratification status of the point cloud.In the statistical analysis of the distance between the corresponding points,both mean and variance are better than the results of precise registration.4.Based on RANSAC algorithm,the method of evaluating the quality of point cloud was designed.The theoretical model of point cloud was established by RANSAC algorithm.By measuring the distance of the actual point cloud to the theoretical model and the data processing,the error distribution of point cloud relative to the theoretical model was got,and then the actual quality of the point cloud was evaluated.The quality evaluation data shows that the technical route used in this paper can effectively improve the accuracy of the surface topography characteristics of the point cloud. |
PDF Full Text Request