Skip to content

Latest commit

 

History

History
217 lines (167 loc) · 14.2 KB

README_borea_lib.md

File metadata and controls

217 lines (167 loc) · 14.2 KB
Raw

IGNF badge PyPI Downloads

Borea is an open-source python package tools-box photogrammetric conversion format and transformation coordinate of image and terrain.
Why Borea? B for Box and orea is a back slang of aero.

Installation

Pip

Due to different dependency used installation of the library require GDAL>=3.3.2, which is not included in the dependency.

pip install ign-borea

For GDAL installation you need libgdal-dev:

sudo apt-get install libgdal-dev

Please note that the GDAL version depends on the libgdal-dev version.

apt-cache show libgdal-dev
# or if you are ogr
ogrinfo --version

pip install GDAL==<GDAL VERSION FROM OGRINFO>

You can find more information on mothergeo-py if you have problems installing GDAL.

In the QGIS environment

View the doc on borea github docs/installation/In_QGIS.md.

Tools

Some tools are already implemented in the library:

  • Conversion OPK to OPK: opk-to-opk -h doc (OPK = Omega Phi Kappa)
  • Control OPK file: opk-control -h doc
  • Conversion OPK to RPC: opk-to-rpc -h doc
  • Conversion OPK to Conl: opk-to-conl -h doc (Conl = light conical file, IGN France format)
  • Transforms coordinates terrain from image: pt-image-to-world -h doc
  • Transforms coordinates image from terrain: pt-world-to-image -h doc
  • Transforms coordinates file terrain from image: ptfile-image-to-world -h doc
  • Transforms coordinates file image from terrain: ptfile-world-to-image -h doc
  • Calculates opk by space resection: spaceresection-opk -h doc
  • Transform projection of points file: transform-proj-points -h doc

Read data and instantiate worksite

Creation of a worksite object from a worksite file (.opk) to be read by reader_orientation(pathfile, arg_dict) with from borea.reader.orientation.manage_reader import reader_orientation.
arg_dict is a dictionary for different args:

  • "interval":[first_line, last_line] is an list of int that specifies the number of lines you want to read. first_line allows you to skip the file header, which must not be taken into account when reading the file, as specified in the header variable. If first_line = None skips everything up to last_line, if lastline = None skips everything from first_line to the end, and if both are None reads the entire file.
  • "header":header described in the section above, is a list of str e.g. ['N', 'X', 'Y', 'Z', 'O', 'P', 'K', 'C'] = list("NXYZOPKC"), detail of letter at section header of file opk below.
  • "unit_angle": "degree" degree or radian.
  • "linear_alteration": True boolean saying True if z shots are corrected by linear alteration.
  • "order_axe: "opk" string to define the order of angle to calculate rotation matrix.

Once the object has been created, you can add other data to it:

  • Setup the projection of the worksite work.set_proj(epsg, path_geoid), with:

    • epsg the code epsg e.g. 2154
    • path_geoid path to the file pyproj GeoTIFF of geoid

  • The camera with read_camera([filepath], worksite) in from borea.reader.reader_camera import read_camera, this function only reads txt and xml files referencing camera data, and can take several camera files if there are several.

  • Link points with read_file_pt(filepath, header, type_point, worksite) in from borea.reader.reader_point import read_file_pt. this function reads all .txt, .mes, .app and other file types, as long as the data structure in the file is column-based and delimited by spaces. The first args is the file path of one file. The second is the column type in the file detail of letter at section header of point file below. The third is the type of point 'co_points' for connecting points, 'gcp2d' for coordinnate of gcp in images and 'gcp3d' for gcp coordinate in the ground. And the last args is the worksite where data will be save.

  • Add Dtm to your worksite work.set_dtm(path_dtm, type_dtm), It converts z data between gcp and acquisition position if these are not in the same unit (one in altitude and one in height). type_dtm is the unit of the dtm 'altitude' or 'height'.

Examples in section examples below.

Different process

  • Set different parameters for shots (projection system of shot and z_nadir), mandatory if data is to be processed afterwards. work.set_param_shot().

  • Can calculate the position of image points in world with ImageWorldWork(worksite).manage_image_world(type_point, type_process, type_control) in from borea.transform_world_image.transform_worksite.image_world_work import ImageWorldWork.

    • type_point is the type of point you want to calcule co_points or gcp2d.
    • type_process is the type of process you want to use intersection with key inter or least square methode with key square.
    • type_control egal None by default, is used if the type_point = gcp2d and if you want just one type code point, else None to process on all point.

    The result can be found in worksite.co_pts_world['name_point'] when type_point = co_points or worksite.img_pts_world['name_point] when type_point = gcp2d.

  • Can calculate the position of terrain points in images with WorldImageWork(work).calculate_world_to_image(type_control) in from borea.transform_world_image.transform_worksite.world_image_work import WorldImageWork, with type_control egal None by default, is used if the type_point = gcp2d and if you want just one type code point, else None to process on all point. . The result can be found in worksite.shots['name_shot'].gcps['name_gcp'] for each image and each gcps.

  • Can calculate spatial resection for each shot in worksite with SpaceResection(work).space_resection_on_worksite(add_pixel = (0,0)) in from borea.transform_world_image.transform_worksite.space_resection import SpaceResection. add_pixel is used to add a mainiation to the position of the points to modify the shot's 6 external parameters for data conversion.

  • Can calculate spatial resection in poitn of shot for creating worksite with SpaceResection(work).space_resection_to_worksite(pt2d, pt3d, pinit) in from borea.transform_world_image.transform_worksite.space_resection import SpaceResection.
    The DataFrame pt2d is a table with 4 column and n line. The id of column must be:

    • id_pt: the id of the point
    • id_shot: the name of the shot where the point is located
    • column: column coordinate in pixel of the point in the image
    • line: line coordinate in pixel of the point in the image

    it can be created with the function read_file_pt_dataframe(path_file_pt,header_file,"pt2d")
    The DataFrame pt3d is a table with 5 column and n line. The id of column must be:

    • id_pt: the id of the point
    • type: if point is gcp with type else None
    • x: x coordinate in your projection system of the point
    • y: y coordinate in your projection system of the point
    • z: z coordinate in your projection system of the point

    it can be created with the function read_file_pt_dataframe(path_file_pt,header_file,"pt3d")
    The dictionary pinit which give the initialization point X, Y, Z. A point on the worksite with a z at an approximate flying height. The name of the key in the dictionary is coor_init.
    Example at the end of explanation of function file.

  • You can calculate some control point statistics to see how accurate your site is stat = Stat(work, pathreturn, control_type) to init the object and run for all stat with stat.main_stat_and_save(). Make stat on function image to world and world to image, if there are data. And save result on pathreturn/Stat_{Name_worksite}.txt.

Examples in section examples below.

Write data

  • Can write worksite object as different format OPK, RPC, Conical for GEOVIEW. The function is manager_reader(writer, name, pathreturn, args, work) in from borea.writer.manage_writer import manager_writer:
    • writer (str), is the type of output "opk", "rpc", "con".
    • name (str), name of file to save, just to save in opk, for other format this args isn't read.
    • pathreturn (str), path of folder where you want to save data.
    • args (dict), Dictionary with different args for the format to save, detail at setion args for writing file below.
    • work (Worksite), the worksite to save.

Examples in section examples below.

Examples

All examples are in borea github ./examples/:

  • For build main class Worksite with file eg_build_worksite_by_file.py and with data eg_build_worksite_by_data.py.
  • To make transformation image to world eg_image_to_world.py.
  • To make transformation world to image eg_world_to_image.py.
  • To make space resection on point to determine worksite eg_space_resection.py.
  • To convert format opk to an other format opk rpc con eg_opk_to_format.py.

Examples of the different formats of file can be found in borea github ./dataset/:

  • An opk file 23FD1305_alt_test.OPK with z unit is altitude.
  • Cameras filesformat Camera1.txt and Camera2.txt.
  • Geotiff of the French geoid for pyproj fr_ign_RAF20.tif** detail at section info projection below.
  • Crops geotiff of the French DTM MNT_France_25m_h_crop.tif in height unit.
  • Ground Control Point (GCP) in terrian GCP_test.app unit z in height.
  • Ground Control Point (GCP) in image terrain_test.mes.
  • Connecting points in image liaisons_test.mes.
  • Image point to transform terrain coordinates to image coordinates to find out in which image the points are located terrain_test0.mes.

Detail

Header of file opk

header is used to describe the format of the opk file read. It provides information on what's in each column, and gives the data unit for Z and angles.
Type is:

Symbol Details
S to ignore the column
N name of shot
X coordinate x of the shot position
Y coordinate y of the shot position
Z coordinate z altitude of the shot position
H coordinate z height of the shot position
O omega rotation angle
P phi rotation angle
K kappa rotation angle
C name of the camera

Header of point file

header is used to describe the format of the point file read. It provides information on what's in each column.
Type is:

Symbol Details
S to ignore the column
P name of the point
N name of shot
T type of point
X coordinate x of the shot position
Y coordinate y of the shot position
Z coordinate z altitude of the shot position

Camera file format

The camera file is a txt file, containing 6 pieces of information about the camera : its name (str), ppax (float), ppay (float), focal (float), image size: width (int) and height (int) in pixels.
Ppax and ppay are the main points of image deformation in x and y directions.
Each line of the file corresponds to a piece of information, starting with the type = info.

name = UCE-M3-f120-s06
ppax = 13210.00
ppay = 8502.00
focal = 30975.00
width = 26460
height = 17004

Only these 7 pieces of information will be read. You can add comments with a # in the first element of the line or other type = info, but they will not be read by the tool, unless the attribute has been added to the Camera class in borea/datastruct/camera.py. An example file can be found in repository ./dataset/Camera1.txt.
No camera-related distortion is taken into account (distortion-free camera).

Info projection

This library can transform and process 3D data with a z in altitude or height. This is done by the pyproj library, which needs the geoid at site level to change units.

The varaible in example for adding a geoid is path_geoid, a list which contains paths of geoids, where you can enter the paths to the various geoids. If the file is stored in pyproj's native folder (pyproj.datadir.get_data_dir(), usr/share/proj or env_name_folder/lib/python3.10/site-packages/pyproj/proj_dir/share/proj) the file name is sufficient pyproj will find it on its own. Geoids file can be found on pyproj's github (https://github.com/OSGeo/PROJ-data).

Args for writing file

OPK

There are 4 keys in the dictionary:

  • "order_axe" (str): Order of rotation matrix axes.
  • "header" (list): List of column type file (same to read opk).
  • "unit_angle" (str): Unit of angle 'degree' or 'radian'.
  • "linear_alteration" (bool): True if data corrected by linear alteration or else False.

RPC

There are 3 keys in the dictionary:

  • "size_grid" (int): size of the grip to calcule rpc.
  • "order" (int): order of the polynome of the rpc. [1, 2, 3]
  • "fact_rpc" (float): rpc factor for world coordinate when not src, we recommend None.

CON

There is no need for an additional argument, you can set None to the argument.

logo ign logo fr