GitLab

[FIX] Use osg::Matrixf instead of osg::Matrix

[FIX] fiber stipples finally work now; there might still be a problem with the texture lookup at the voxel positions

[FIX] fiber stipples now actually render stipples again; there is still a problem with stipple directions

Previous commitment had some bug fixes. But it also said that some bugs dealing
with the surface segmentation still remained. This commit fixes them. The
errors consisted of not initialized member variables in some classes regarding
point instances of parameter space clustering and boundary detection.

Regarding the module "Surface detection by Lari" crashes appeared sometimes.
The reason was that the object containing the status bar thing (point classifier
class) was deleted before the progress was set to be finished.

Remaining bug:
The following bug could not be fixed short-time because it does not appear if
the module is compiled using the debug flag.
Shortly said sometimes the search instance even can not find a point using
coordinates of itself. Weird. I put setting the ID in the end of
WLariBruteforceClustering::addExtentCluster manually at least not to get stuck.
But still some groups result a bad result. No errors appear when compiling (cmake)
using the debug flag. Has to be fixed sometime later.

Having forgotten to commit memory bugs previously. Changes are commited here.

Main changes:
  - Multithreading for allocating kd-trees at 8 threads.
  - Fixed memory leak when kd-tree is destroyed. Not all memory was freed previously.
  - Added few point transformation features regarding automatical color adjustment.
  - Improved speed of the point file lading feature.
  - Fixed ReadLAS module
      - Previously it could only use greyscale color channel.
      - Made X/Y-sliders more intuitive

 * Implemented status bar feature for "Surface detection by Lari/Habib" and "Point Groups - Validator"

 *  - Just completed and fixed up things
 *  - Fixed bug contained in "Point groups - Transform". Originally there was a
 *    performance trouble during processing point data with very many groups.

 * Added some empty lines between methods in code and header files.
 *
 * Bugfix: Solved problem in point data saving that the coordinate precision was about 5
 * signs.
 *
 * Sebastian Eichelbaum wrote (18.09.2014, 11:46):
 * >> Hi Andreas
 * >>
 * >> Ich würde dich gern bitten darauf zu achten, dass du zwischen Funktionen
 * >> eine Leerzeile lässt. Das betrifft sowohl die CPP als auch die Header
 * >> files. Als Beispiel: WBmpImage.cpp. Dort hast du nach dem Konstruktor
 * >> eine Leerzeile gemacht, nach dem Destruktor und den folgenden Funktionen
 * >> jedoch nicht.
 * >>
 * >> Ansonsten ist dein Stil sehr sauber. Auch die Doxygen Doku ist sehr
 * >> ordentlich. Super.
 * >>
 * >> Gruß
 * >> Sebastian

 * Modules purged
 *   - Points group selector
 *
 * Altered modules
 *   - Points - Transform: Can now save and load point data using files
 *
 * Modules created:
 *   - Point Groups - Validator
 *       - Is done in order to validate groups against reference groups
 *   - Point Groups - Transform
 *       - Can save and load grouped point data using files
 *       - Also can merge grouped point data sets and subtract point sets
 *       - Still has all features to replace "Point group selector" completely

 * General style Some style improvements.
 *
 * General features:
 *   - Points - Transform: Color equalizer (contrast and offset setting for each color
 *     channel)
 *
 * Performance improvements (in general about 4 times faster)
 *   - Lari/Habib 2014 clustering has now following improvements.
 *       - Removing points from the parameter space kd-tree after adding to a plane -
 *         This change created an additional, temporary parameter domain kd-tree which
 *         is in the beginning just a copy of the original parameter domain kd-tree.
 *         When points are added to a plane, they are removed from that copy. It makes
 *         the algorithm a way faster.
 *       - Distinguishing whether a point belongs to an extent directly in an point
 *         searcher instance works faster than building a list with additional
 *         unnecessary points and filtering afterwards. In that change a new class
 *         inherited from WPointSearcher and overwrote a method which distinguishes
 *         whether points are added or not.
 *       - Another story is counting points of an extent with a searched point as peak
 *         centre. Previously points were put in a list in order to count them
 *         afterwards. This new class was further modified at traversing. So no new
 *         unnecessary list is built up.
 *
 * Bugs fixed:
 *   - Boundary detection problems were fixed. Originally there was a detection that
 *     checked whether a new bound would hit the pre previously detectet bound. But that
 *     seemet practically not to be enough. So every point adding a new bound is checked
 *     whether it hits one of all bound pieces detected. Nevertheless the problem is not
 *     solved completely but there are much less less problems by far.

 * Previously application crashed every time on processing data using Lari/Habib (2014) surface detection
 *
 * In method WMSurfaceDetectionByLari::properties():
 * WPropInt m_cpuThreadCount = m_properties->addProperty( "CPU threads: ", "", 8, m_propCondition );
 *
 * Yes, m_cpuThreadCount is a member variable of the class. Putting "WPropInt" here is
 * definitely not a good idea.

 * Finalized main features
 *
 * New added features:
 *   - Brute Force peak detection clustering
 *   - Modified convex hull boundary detection for separation of spatially disconnected
 *     points.
 *
 * Reached beta version status.

 * Finalized main features
 *
 * New added features:
 *   - Brute Force peak detection clustering
 *   - Modified convex hull boundary detection for separation of spatially disconnected
 *     points.
 *
 * Reached beta version status.

 * Finalized main features
 *
 * New added features:
 *   - Brute Force peak detection clustering
 *   - Modified convex hull boundary detection for separation of spatially disconnected
 *     points.
 *
 * Reached beta version status.

 * Finalized main features
 *
 * New added features:
 *   - Brute Force peak detection clustering
 *   - Modified convex hull boundary detection for separation of spatially disconnected
 *     points.
 *
 * Reached beta version status.

 * Steps done:
 *   - Implemented the plane point clustering approach of Lari/Habib
 *
 * Probable remaining steps that remain to be implemented:
 *   - Applying a modified convex hull boundary detection algorithm to separate
 *     Spatially disconnected planar patches that were detected to belong to the same
 *     Planar formula.

[ADD #371] Lari/Habib approach module can display point's best fitted planes and the parameter domain
 * Added features:
 *   - Displaying the best for each pointfitted plane (calculated using the least
 *     squares method in relation th neighbor points)
 *   - The parameter domain points can be displayed now. Each oof them displays the best
 *   - fitted plane properties
 *
 * Bug fix:
 *   - Fixed the Hessesche normal formula calculation. Previously A, B and C were
 *     normalized just like a normal vector (Actually they depict a normal vector)

 * First module: Creates a sphere or cube of random points of the radius of 2.
 * Second module: Visualizes the least squares process by putting the output plane to
 * the triangle mesh output.
 * Transformed the points cropping module in order to be able to stretch and rotate the
 * point set

 * First module: Creates a sphere or cube of random points of the radius of 2.
 * Second module: Visualizes the least squares process by putting the output plane to
 * the triangle mesh output.
 * Transformed the points cropping module in order to be able to stretch and rotate the
 * point set.

[FIX #373] Due to the update change of #373 there was a minor update bug: When new data was connected, no update took place.

 * This approach describes the surface detection of Zahra Lari and Ayman Habib (2014).
 * Paper: "An adaptive approach for the segmentation and extraction of planar
 * andlinear/cylindrical features from laser scanning data", ISPRS Journal of
 * Photogrammetry and Remote Sensing, 2014
 *
 * Eventually the process covers also the linear and cylindrical feature segmentation
 * but in our requirements only the planar segmentation is required. Steps:
 *   - Original laser point cloud as input data
 *   - PCA-based classification of planar features
 *   - Selection of appropriate representation models for the detected planar features
 *   - Local point density estimation along the detected planar features
 *   - Precise estimation of the planar features segmentation attributes
 *   - Parameter-domain segmentation and extraction of planar features
 *   - Boundary tracking for resolving the parameter-domain segmentation ambiguities of
 *     planar features
 *   - Extracted planar features in spatial domain
 *
 * Estimations (Based on the paper because I haven't completely impelemented yet):
 *   - Fair segmentation results
 *   - Relatively slow
 *
 * Implementation progress:
 *   - Done to the step "Precise estimation of the planar features segmentation
 *     attributes".
 *
 *
 * Other updates:
 *   - Renamed module "Pooints - Crop" to "Points - Transform". It optained the feature
 *     to translate and to rotate the point set.
 *   - Wew code refactoring.