1, the principle of three-dimensional reconstruction
Point cloud data, through deep image enhancement, point cloud calculation and registration, data fusion, surface generation and other steps to complete the three-dimensional reconstruction of the scene. In depth image acquisition, in order to obtain enough images, it is necessary to change different angles to capture the same scene to ensure that all the information of the scene is included [3]. The specific solution can be either a fixed camera to capture an object on the rotating platform, or a rotating camera to capture a fixed object. Then the image enhancement process such as denoising and repairing the depth image is performed. The preprocessed depth image has two-dimensional information, and the value of the pixel is depth information, indicating the linear distance between the surface of the object and the camera. For multi-frame scene images taken at different angles, each frame contains a certain common part, which requires point cloud registration processing to eliminate redundant information. The point cloud data is then fused to obtain a more refined reconstruction model, and the visible isosurface of the constructed object is generated through the surface.
Principle of 3D reconstruction
2, shooting : large objects, such as sculptures, buildings, shooting by means of drones flying around. A small object, using a mobile phone or a camera, to shoot around a target object, the shooting method is referred to the drone. Small object modeling, sometimes you can place an object on a rotatable stool for shooting.
UAV live shooting method:
Operate the drone to fly over the object and raise it to a suitable height to see the whole and turn on the video recording function. First take a 30S shot. Then adjust the drone to the point of interest to fly around, set the flying radius to 50M, and set the aircraft speed to 5M/S. After flying around a circle, lower the height to half, adjust the lens to the shooting target, and then fly around.
3. Copy the captured video to the computer and use the PR software to export the video to the video in a single frame to ensure that each photo has a 30% repeatability. Take 4 photos in a low-pitched manner, and fly 2 circles of 8 photos per circle, totaling 20 photos.
(1) New PR project.
(2) Right-click to import the media file.
(3) Drag the file onto the timeline. Then click to play the selected photo frame.
(4) Save the photo frame single frame export as a photo.
3, altizure reconstruction: import into the cloud platform altizure, after photo alignment, build dense point clouds, generate mesh, generate textures, you can quickly generate 3D models.
(1) Open the altizure website. Sign up for free and click on New Project.
(2) Enter the project title and click Next.
(3) Add an image set and click Next.
(4) The system automatically calculates and reconstructs automatically. After the completion, the interface is as follows. The general project takes 20 minutes. After the project is completed, copy the project link and send it to WeChat. It's easy to share.
4, using PhotoScan for 3D reconstruction: accuracy and speed are inferior to the altizure cloud platform, but PhotoScan is installed and run on a PC, which can easily generate the required files and even print with a 3D printer.
(1) Open the file and add the heap block.
(2) Select the heap block, right click to add a photo, then align the photo, select the "high precision" option, do not select the highest, it takes too long.
(3) Follow the steps to select the dense point cloud, generate the mesh and generate the texture, and finally form the project display.
(4) Specific operational details, you can ask the mother. Here only to throw bricks and jade.
SHENZHEN CHONDEKUAI TECHNOLOGY CO.LTD , https://www.szsiheyi.com