You are here: Start » Program Examples » Calibration with filters on remaped image

Calibration with filters on remaped image

Aim

The goal is to compute the world coordinates based on given images of the calibration grid. The application uses filters instead of the editor.

Input

A few images of the calibration grid and a board with circles.

Output

The coordinates of the found dots.

Hints

First of all, as there are both lens and perspective distortion, they need to be removed by using the RectifyImage filter. Next, to correctly locate the dots it is best to use the ExtractBlobs_Intensity. Eventually, to find the center points of circles, try fitting a corresponding shape to them and use an appropriate property output.

Solution (AVS)

  1. In the INITIALIZE section create a Step macrofilter and name it Calibration.

  2. Inside the Calibration macrofilter create a Task macrofilter and name it FindCameraCalibrationGrids. Create a new input of the Task o type Directory. Set its value to the directory to the Calibration Grid images.

  3. Inside the FindCameraCalibrationGrids macrofilter:

  4. Return to the Calibration macrofilter. Add the CalibrateCamera_Pinhole filter. Connect its inputs with the FindCameraCalibrationGrids outputs.

  5. Create a new Step macrofilter named FindWorldCalibrationGrid. Drag the outCameraModel output of the previous filter to the new Step to create the input.

  6. Inside the FindWorldCalibrationGrid macrofilter:

  7. Return to the Calibration macrofilter. Add the CreateRectificationMap_PixelUnits filter:

    • For the inImageFormat input connect the Width and the Height to the Width and the Height outputs of the FindCameraCalibrationGrids macrofilter.
    • Connect the inTransform input to the FindWorldCalibrationGrid output.
    • Drag the outRectificationMap and the outRectificationMap.RectifiedWorldScale outputs to the Macrofilter Outputs section.

  8. Return to the Main macrofilter. Add the EnumerateImages and specify the directory to images in the inDirectory.

  9. Add the RectifyImage filter. Connect the outImage to the inImage. Connect the inRectificationMap input to the RectificationMap output of the Calibration macrofilter.

  10. Create a new Step macrofilter named FindDiscs. Drag the RectifyImage outImage output to the new macrofilter to create the input.

  11. Go to the FindDiscs macrofilter. Add the ExtractBlobs_Intensity filter. Connect its inImage input to the macrofilter's input.

  12. Add 2 ClassifyRegions filters:

  13. Add the RegionBoundingCircle_OrNil filter and connect its input to the last ClassifyRegions outAccepted output.

  14. Add the FitCircleToEdges filter:

  15. Add a formula. Drag the outAlignedFittingField Axis.Radius out put form the previous filter to create the input and name it Radius.

    outDiameter = inRadius * 2

  16. Drag the formula output to the Macrofilter Outputs section.

  17. Go back to the Main macrofilter. Add the CreateCoordinateSystemFromPoint filter. Set the inPoint input as {50;50}.

  18. Add the AlignPoint filter. Connect its inAlignment input to the CreateCoordinateSystemFromPoint output and the inPoint input to the ImageWasherCenter output of the FindDiscs macrofilter.

  19. Create a new Step macrofilter named PixToMilimeters. Drag to it:

    • the AlignPoint output
    • the Calibration macrofilter RectificationMapRectifiedWorldScale output and name the new input as WorldScale
    • the Diameter output of the FindDiscs macrofilter.

  20. Inside the PixToMilimeters macrofilter:

    • Label the WorldScale input.
    • Exapnd the AlignedPoint input to get X and Y values separately.
    • Add 3 PixelsToMillimetres filters:
      • Connect the inScale input with the WorldScale for each new filter.
      • For the first one, connect the inPixels to the Diameter macrofilter's input.
      • For the second one, connect the inPixels to the X macrofilter's input.
      • For the third one, connect the inPixels to the Y macrofilter's input.
    • Connect each PixelsToMillimetres outputs to the macrofilter's output and name them: Diameter_mm, PositionX_mm and PositionY_mm.

  21. Go back to the Main macrofilter. Create a new Step macrofilter named GenerateGraphics. Drag to it:

  22. Inside the GenerateGraphics create 2 Step macrofilters and name them VisualizeCordinateSystem and CreateLabels. Drag the Alignment macrofilter input to the VisualizeCordinateSystem step and all the rest of the inputs to the CreateLabels.

  23. Inside the VisualizeCordinateSystem macrofilter create a new Step macrofilter. Name it GeneratingAxis. Create 3 inputs of it:

  24. Inside the GeneratingAxis macrofilter:

    • Add the MakeSegment filter. Connect the inPoint2 input to the macrofilter's Point2 input.
    • Add the AlignSegment filter. Connect the inSegment input to the previous filter's output. Connect the inAlignment input to the macrofilter's Alignment input. Expand the outputs to get the outAlignedSegment Point2 value. Drag the outAlignedSegment output to the Macrofilter Outputs section.
    • Add the MakeStringLabel filter. Connect the inText input to the macrofilter's Axis input. Connect the inPosition input to the Point2 output of the previous filter. Drag the outLabel output to the Macrofilter Outputs section to create a new output.

Macrofilter Main.

Macrofilter Calibration.

Macrofilter CreateLabels.

Macrofilter FindCameraCalibrationGrids.

Macrofilter FindDiscs.

Macrofilter FindWorldCalibrationGrid.

Macrofilter GenerateGraphics.

Macrofilter GeneratingAxis.

Macrofilter PixToMilimeters.

Macrofilter VisualizeCordinateSystem.