DSM Quality Assessment


Rasterization

During the DSM process SURE will map points of the calculated point clouds into bins (pixels/raster cells) to generate a DSM. Besides the actual result of the rasterization, the following images, containing additional information for each cell, can be generated by SURE and stored in separate folders inside the DSM folder:

HISAn 8bit image with "height ramp" visualization style for the DSM
DSM_StdDevA 32bit image, containing the standard deviation for each pixel, calculated from a square mask (default: 3x3).
DSM_ModelCountAn 8bit image, containing the number of stereo models involved in calculating the height.
NUMAn 8bit image, containing the number of individual 3D points detected per raster cell
NUMVISAn 8bit image - the scaled version of NUM (in grey values), for visualization purposes
STDA 32bit image, containing the standard deviation for each pixel, calculated for the population of 3D points contained within the raster cell

Interpolation

During the rasterization, it can happen that some of the pixels of the resulting DSM remain empty (NaN). In order to fill those pixels SURE relies on interpolation. In order to enable users to assess the reliability of the interpolated DSM and derived products like True Orthos easily, SURE offers additionally the following images, providing relevant information about the interpolated raster cells:

DSM_BinaryMaskAn 8bit binary mask, containing 0 for every interpolated pixel and 255 for every measured pixel.
DSM_DistanceMaskA 32bit distance mask, containing the euclidean distance to the nearest measured pixel.

How to..

..generate the images from the SURE UI:

  1. Go to: Advanced Configuration → DSM
  2. You will find four checkboxes to enable binary mask, distance mask, standard deviation and model count
  3. Use the mask size parameter to determine the size of the square mask you would like to be used for the calculation of the standard deviation

..generate the images from the command line:

  1. The sure command line offers four flags that can be used to enable the generation during the DSM process: dsm_binaryMask, dsm_distanceMask, dsm_stdDev, dsm_modelCount
  2. The mask size for calculating the standard deviation can be changed inside the controlDsm.txt (see next point)


If you want to generate the project structure including the control files without running SURE you can use the "-prep" flag in the command line. This will enable you to modify control files before running SURE.


..enable the generation from controlDsm.txt:

  1. Inside the controlDsm.txt of your current project you will find the following settings: $MakeHeightImageVis (HIS), $MakeStdDevImage (DSM_StdDev), $StdDev_MaskSize (size of window containing the population of pixels for which the standard deviation - DSM_StdDev - is computed), $MakeModelCountImage (DSM_ModelCount), $MakeCountImage (NUM), $MakeCountImageVis (NUMVIS), $MakeCellStdImage (STD), $MakeBinaryMaskImage (DSM_BinaryMask), $MakeDistanceMaskImage (DSM_DistanceMask).  
  2. Except for the $StdDev_MaskSize, the other parameters are simple on/off switches, set them to 0 to turn the generation off and to 1 to turn it on.
  3. The $StdDev_MaskSize parameter can be used to control the size of the square mask used during the calculation of the standard deviation image.