Seed Grown Match

From fokus-gmbh-leipzig.de
(Redirected from SGM)
Jump to: navigation, search

'Seed Grown Matching (abrev. SGM) is a term describing a special matching algorithm which has been created to create point clouds from stereo models. All data is stored in a binary file format (*.EAS). This file contains all information about the SGM and how it was created. This data are

After you have orientated the stereo image pair of a stereo project absolutely. You are asked at the bottom of the absolute orientation tool window to save the model. By pressing the button Save Model... aside to the stereo model a blank EAS-file is saved below the seed grown matching node of the DSM tree.

The context menu of the SGM node

The context menu of a SGMatch

  • Update Orientations and Object Coordinates
  • Remove SeedGrownMatch
Deletes the selected SGM from the model list at the seed grown matches node of the DSM tree but not the .EAS file from project directory.

Kommentare sind nur von angemeldeten Nutzern sichtbar

  • Delete only generated Point Cloud
Deletes only the point data from the .EAS file, but not the selected SGM from the model list.
  • Delete all Data (inclusive Seed Points)
Deletes all data including seed points and images from the .EAS file, but not the selected SGM from the model list and not the .EAS file from project directory.
  • Release used Memory

Releases the memory used by the SGM by deactivating it.

  • Export
  • *.tri ...
  • Coordinates...


  • Coordinates Seed Points...
  • Image Coordinates of the Reference Image...
  • Lupos 3D-Export (*.ptb) ...

  • Coordinates (Test-GFaI)...
  • STL-Export...
  • VRML-Export (Points)...
  • VRML-Export (DSM)...
  • VRML-Export (DSM textured)...

Gives several export options of the generated point cloud:

a binary *.TRI file, a metigo-format,

a coordinate file in predefined formats at the software:

*.PKO, *.BKO, *.PKB, *.TAC;

the seed points as a coordinate file in the formats listed above;


to the Lupos 3D-format with two scale settings;



the triangulated point cloud to a *.STL file;
the textured points of the cloud to the VRML format;
the triangulated point cloud to the VRML format;
the textured triangulated point cloud to the VRML format;
Note Note: You can choose between two export options for Lupos, STL and VRML via the prompted dialogue.

  • Reinitialise local DSM
  • Rename local DSM
  • Delete Triangles
  • Selection
    • Delete Selection
    • Cancel Selection
    • Invert Selection

The toolbox seed grown matching (SGM)

The toolbox seed grown matching

To access the toolbox seed grown matching select a blank SGM-file (*.EAS) from the DSM tree and press the SGM-button 3DIconDSM SGM-active.png from the DSM toolbar.

  1. At the top the name of the currently selected SGM is shown.
  2. At the following section you select the images which you want to use to create the SGM by pressing the button Select Images....
    You can select more than two images but only the top two images in the list will be used as best reference for determing seed points and generating the point cloud. The additional images are used for improving the intersections and therefore the accuracy of the data (points) generated via the reference image pair.
    • If the blank SGM-file (*.EAS) has been generated via a stereo project the stereo image pair will be automatically loaded and used as reference image pair.
    • If you create the SGM via the context menu of the seed grown matches node of the DSM toolbar you have to move the reference image pair to the top of the list by drag'n'drop.
  3. At the text on the right of this section the number of 3D-coordinates calculated by absolute orientation in the stereo project. If the number or location of the seed points is insufficient or none has been created press the button to the left. At the prompted dialogue you can either search for seed points, load them from a 3D-coordinate file or measure them in the stereo window of the stereo model.
  4. At the section Generating Point Clouds you start the SGMatching algorithm or add an already added match and set the parameters with which matching is conducted.
    • The setting Increment [px] is taken from the input field Density of points on the object of the New Project dialogue but can also be set manually. Changing the increment will automatically modify the text field thereafter which displays the density of points on the object.
    • Press the button Matching-Parameters to modify the matching parameters. These are identical to those already described for paragraph 5 and 6 of the extended Automatic Point Search dialogue.
    • To conduct seed grown matching for the first time press Start SGMatching. In the evaluation window you now see how the point cloud in rhombic shape gradually grows around the orange seed points.
      After successful matching the button will turn grey.
    • To add to the point cloud press the button Add SGMatching. This might be done if you add further seed points at areas not or insufficiently matched or altered the matching parameters via the Matching-Parameters button.
  5. Processing a batch of seed grown matches
    You also can create a batch of several SGMs to be run during a break.
    • To create the batch select the blank model entry in the seed grown matches node of the DOM tree and press the button SGMatching in Batch. At the text field to the right you see the number of data sets in the batch.
    • Press the button Process Batch List to run the SGM-algorithm. The Dialog extended Seed Grown Matching (Batch Mode) will be prompted. This dialogue shows a table with the individual models in rows to be processed. The columns display the model's name, the number of images used, the state of processing. The text field at the bottom of the dialogue informs about the model currently processed
      As soon as the of one model in the batch has been processed the number of points created and the time needed are displayed in the state column. Press the button Quit to exit the dialogue and to abort the batch processing.
    • Press the button Clear Batch List to remove all unprocessed SGMs from the batch.
  6. During matching the point cloud will be automatically textured with the respective colour of the first image in the table at paragraph 2. To show the quality of the construed points you can select between different colour codings.
    • You can select from the pull-down-menu one of the following variables and enter its interval in the text field to its right:
      • mp(x,y,z) - the point error of photogrammetric intersectioning as an average of the errors of the three coordinates in [cm];
      • mx - the point error of photogrammetric intersectioning in the x-coordinate in [cm];
      • my - the point error of photogrammetric intersectioning in the y-coordinate in [cm];
      • mz - the point error of photogrammetric intersectioning in the z-coordinate in [cm];
      • dEpi - the distance to the epipolar line in [px];
    • By clicking onto the two colour fields at the very right you can modify the colour for the smallest value first and the greatest value second.
    • Enter the number of colour intervals you want to display.
      At the preview window below the colour range is visualised.
    • To execute the settings above press the button New Colouring.
    • To create a log file for the epipolar geometry press the button Epi Log....
  7. At the bottom section of the tool window you can select low quality points.
    • Press Colour Selection... to select one or more size ranges of the colour definiton as described in section 6.
    • Press Delete Selection... to delete the selected colour range.
    • Press Clear Selection... to clear the selected colour range.

Kommentare sind nur von angemeldeten Nutzern sichtbar

The dialogue Seed Points

The seed point dialogue
This dialogue opens if you have pressed the Seed Points button at section 3 of the SGM tool window.
  1. At the top section of the dialogue you can select among the following options:
    • Press the button Search Seed Points to search for seed points in the stereo model. The parameters for the search are given in section 2 and 3.
    • Press the button Load Seed Points to load any file of 3D-coordinates.
    • Press the button Measure Seed Points to measure a seed point in the stereo model. Within the stereo model go to the required area of the model and press the FP-key, move to the point you want to measure and right-click. A dialogue will be prompted where you confirm that you want to use the point as new seed point.
  2. Parameter Point Search
    • starting depth [4-6]]
    • number of points
    • search raster
    • tolerance [0-2]]
    • search window [4-8]]
  3. For desciption of the Dimensions of Matrices, Thresholds and Search Rangesin this section compare paragraph 5-7 of the extended Automatic Point Search dialogue.

Kommentare sind nur von angemeldeten Nutzern sichtbar

See also