MicMacRoom Tutorial: 05 GrandLeez : Différence entre versions

Version du 24 avril 2024 à 10:54

Wiki Tutorial Sheet - GrandLeez

Dataset Type: Tutorial to illustrate processing on aerial images (drone, airplane, etc.). This pipeline enables the generation of different types of point clouds: one point cloud with relative orientation with respect to the camera of capture and two point clouds with absolute position and orientation for a given reference system.

Data Retrieval: The GrandLeez tutorial data is available on the MicMac wiki ensg. Make sure to download all the necessary data for the processing chain.

Creation of the project

Open the 'File' tab in MicMac and launch a new pipeline with the tamplate "Mic Mac Aero TapiocaFile final". Check if the pipeline is empty. If not, you can erase the old dataset with "Remove All Images". Don't forget to save your new project: MicMac will give to your project a default name that you cannot change. Then drop all your images and dataset on the file of the project, and drop at least one image on the interface.

1) Calculation of Homologous Points: TapiocaFile

Calculation of homologous points in all images. Here, we use the TapiocaFile command to calculate homologous points: the use of this command is made possible by the existence in the dataset of an XML file (FileImagesNeighbour.xml) listing all pairs of overlapping images. With this file, the Tapioca command directly searches for homologous points between these pairs of images, rather than processing all images together (processing time gain). A new pipeline under implementation will allow the use of another variant of tapioca for this tutorial if you do not have an XML file for your dataset.

If your dataset doesn't have this xml file, you can generate it with the command : (A FAIRE)

Required Arguments for TapiocaFile command:

• Project Directory: Absolute path of your project, which provides the path to your dataset folder. Note, your files must be located exactly in this folder, otherwise the command will not be able to find them.
• XML File Path: Path of the XML file of your image pairs. You can fill this argument with the filename if it is located in the project folder entered in the previous argument.
• Resolution: Integer specifying the resolution quality of images in the search for homologous points. The higher the resolution of the images, the longer the processing time for homologous points. Other arguments are optional. If you want to access advanced arguments, check the "advanced" option in "filter attributes". The output files of homologous points will be listed in the HomolTapioca folder.

2) Image Coordinate Transformation:

This step converts the embedded GPS coordinates into a format suitable for MicMac (OriTxtInFile), changes the coordinates of the images into a local system, and calculates the relative speed of the camera (for GPS delay determination). Additionally, you can select a set of central images in the acquisition for camera calibration (this step will be repeated during the various Tapas).

Required Arguments:

• Format Specification: Output specification format (usually OriTxtInFile) of GPS coordinates in the MicMac's own format.
• Orientation File: Orientation file (e.g., in CSV format) that provides for each image in the dataset its coordinates given by the embedded GPS during the acquisition. These are the data that we are trying to transform into the requested specification format.
• Targeted Orientation: Output orientation folder name.
• Change System File: Coordinate transformation file. This string must be of the form: strgSyst1@fileSyst2. "strgSyst1" indicates the original coordinate system of the dataset, and "fileSyst2" indicates the coordinate transformation file into the desired system. The @ symbol indicates the transition from one system to another.

Other Parameters:

• MTD1=1 and CalcV=1 are used to determine the relative speed of the camera (for GPS delay determination).
• Image Center Nb Adjacent Images: Path of the central image of the acquisition and the number of images to take around this central image to calculate the internal calibration of the camera.
• Delay: GPS delay if determined beforehand.

3) Camera Calibration:

Command to obtain the device calibration. To do this, we can use only a portion of the dataset to reduce calculation times (usually central image block in the acquisition).

Required Arguments:

• Calibration Model: Camera calibration model.
• Image Pattern: Folder or set of images to be used for calibration.
• Output Name: Output calibration folder name.

Tapas_2: Relative orientation of images CenterBascule_1: Absolute orientation of images

Inputs:

• ARG1-mandatory: (string) folder/set of images to be oriented
• ARG2-mandatory: (string) folder of all relative orientations output by the Tapas_2 command
• ARG3-mandatory: (string) folder of coordinate change between absolute and relative system, transformation performed by the OriConvert command; for each image, conversion file to switch from absolute to relative orientation.
• ARG4-mandatory: (string) output folder name

Named Args:

• [Name=L1] bool :: {L1 minimization vs L2; (Def=false)}
• [Name=CalcV] bool :: {Use speed to estimate time delay (Def=false)}
• [Name=ForceVert] REAL :: {Weight for forcing Axe of camera to vertical}