Research On Depth Of The Camera For Cockpit3D Reconstruction Technology

Considering the needs of the cockpit3D modeling around the Light-weight flight simulator,aiming at the disadvantages of existing3D reconstruction methods, this paper uses a novel depthcamera (Kinect) and proposes a framework for rapid cockpit reconstruction. Firstly, using the visualfeatures combined with color images achieve data frames coarse alignment; Secondly, accuratelyalign point cloud under different views to point cloud model with depth information; finallytriangulate the point cloud to highly accurate cockpit instrument panel mesh model. The maincontents are follows:1. Considering poor quality of output original depth information and multi-hole problems fromKinect, this paper presents a method based on regional growth and improved bilateral filteringalgorithms after the research on existing filtering algorithms. The experiment results show that thismethod can effectively eliminate holes and random noise on the cockpit instrument panel.2. Using SIFT feature to extract RGB image texture features achieve data frames coarsealignment; Using the random sample consensus (RANSAC) algorithm to remove wrong-matchingfeature points; using depth information and mapping2D data to3D as the initial state of ICPalgorithm; We propose the RGB-D ICP algorithm based feature matching; Experimental results showthat: RGB-D ICP algorithm can effectively achieve accurate registration under different perspectives.3. Using Delaunay triangulation algorithm based on discrete point cloud to3D mesh model data;In the Visual Studio platform, using OpenNI framework, PCL point cloud library, OpenGL, OpenCVlibrary to generate3D visual interface; using open source software Mesh Lab, we can edit models.Finally, considering the reconstruction accuracy and time-consuming, we make the quantitativeanalysis for the3D reconstruction of cockpit; the results show that the proposed method can simplyand quickly reconstruct a high-precision3D cockpit model.