TIndexKernel::FileInfo TIndexKernel::getFileInfo(KernelFactory& factory,
const std::string& filename)
{
FileInfo fileInfo;
PipelineManager manager;
manager.commonOptions() = m_manager.commonOptions();
manager.stageOptions() = m_manager.stageOptions();
// Need to make sure options get set.
Stage& reader = manager.makeReader(filename, "");
if (m_fastBoundary)
{
QuickInfo qi = reader.preview();
std::stringstream polygon;
polygon << "POLYGON ((";
polygon << qi.m_bounds.minx << " " << qi.m_bounds.miny;
polygon << ", " << qi.m_bounds.maxx << " " << qi.m_bounds.miny;
polygon << ", " << qi.m_bounds.maxx << " " << qi.m_bounds.maxy;
polygon << ", " << qi.m_bounds.minx << " " << qi.m_bounds.maxy;
polygon << ", " << qi.m_bounds.minx << " " << qi.m_bounds.miny;
polygon << "))";
fileInfo.m_boundary = polygon.str();
if (!qi.m_srs.empty())
fileInfo.m_srs = qi.m_srs.getWKT();
}
else
{
Stage& hexer = manager.makeFilter("filters.hexbin", reader);
PointTable table;
hexer.prepare(table);
PointViewSet set = hexer.execute(table);
MetadataNode m = table.metadata();
m = m.findChild("filters.hexbin:boundary");
fileInfo.m_boundary = m.value();
PointViewPtr v = *set.begin();
if (!v->spatialReference().empty())
fileInfo.m_srs = v->spatialReference().getWKT();
}
FileUtils::fileTimes(filename, &fileInfo.m_ctime, &fileInfo.m_mtime);
fileInfo.m_filename = filename;
return fileInfo;
}
PointViewSet MergeFilter::run(PointViewPtr in)
{
PointViewSet viewSet;
// If the SRS of all the point views aren't the same, print a warning
// unless we're explicitly overriding the SRS.
if (getSpatialReference().empty() &&
(in->spatialReference() != m_view->spatialReference()))
log()->get(LogLevel::Warning) << getName() << ": merging points "
"with inconsistent spatial references." << std::endl;
m_view->append(*in.get());
viewSet.insert(m_view);
return viewSet;
}
// Make sure that spatialreference works for random readers
TEST(json, issue_2159)
{
class XReader : public Reader
{
std::string getName() const
{ return "readers.x"; }
virtual void addDimensions(PointLayoutPtr layout)
{
using namespace Dimension;
layout->registerDims( { Id::X, Id::Y, Id::Z } );
}
virtual point_count_t read(PointViewPtr v, point_count_t count)
{
using namespace Dimension;
for (PointId idx = 0; idx < count; ++idx)
{
v->setField(Id::X, idx, idx);
v->setField(Id::Y, idx, 10 * idx);
v->setField(Id::Z, idx, 1.152);
}
return count;
}
};
XReader xr;
Options rOpts;
rOpts.add("count", "1000");
rOpts.add("spatialreference", "EPSG:4326");
xr.setOptions(rOpts);
StatsFilter f;
f.setInput(xr);
PointTable t;
f.prepare(t);
PointViewSet s = f.execute(t);
PointViewPtr v = *(s.begin());
SpatialReference srs = v->spatialReference();
EXPECT_EQ(srs, SpatialReference("EPSG:4326"));
}
TEST(LasWriterTest, fix1063_1064_1065)
{
std::string outfile = Support::temppath("out.las");
std::string infile = Support::datapath("las/test1_4.las");
FileUtils::deleteFile(outfile);
std::string cmd = "pdal translate --writers.las.forward=all "
"--writers.las.a_srs=\"EPSG:4326\" " + infile + " " + outfile;
std::string output;
Utils::run_shell_command(Support::binpath(cmd), output);
Options o;
o.add("filename", outfile);
LasReader r;
r.setOptions(o);
PointTable t;
r.prepare(t);
PointViewSet s = r.execute(t);
EXPECT_EQ(s.size(), 1u);
PointViewPtr v = *s.begin();
EXPECT_EQ(v->size(), 1000u);
// https://github.com/PDAL/PDAL/issues/1063
for (PointId idx = 0; idx < v->size(); ++idx)
EXPECT_EQ(8, v->getFieldAs<int>(Dimension::Id::ClassFlags, idx));
// https://github.com/PDAL/PDAL/issues/1064
MetadataNode m = r.getMetadata();
m = m.findChild("global_encoding");
EXPECT_EQ(17, m.value<int>());
// https://github.com/PDAL/PDAL/issues/1065
SpatialReference ref = v->spatialReference();
std::string wkt = "GEOGCS[\"WGS 84\",DATUM[\"WGS_1984\",SPHEROID[\"WGS 84\",6378137,298.257223563,AUTHORITY[\"EPSG\",\"7030\"]],AUTHORITY[\"EPSG\",\"6326\"]],PRIMEM[\"Greenwich\",0,AUTHORITY[\"EPSG\",\"8901\"]],UNIT[\"degree\",0.0174532925199433,AUTHORITY[\"EPSG\",\"9122\"]],AUTHORITY[\"EPSG\",\"4326\"]]";
EXPECT_EQ(ref.getWKT(), wkt);
}
std::vector<PointId> SMRFilter::processGround(PointViewPtr view)
{
log()->get(LogLevel::Info) << "processGround: Running SMRF...\n";
// The algorithm consists of four conceptually distinct stages. The first is
// the creation of the minimum surface (ZImin). The second is the processing
// of the minimum surface, in which grid cells from the raster are
// identified as either containing bare earth (BE) or objects (OBJ). This
// second stage represents the heart of the algorithm. The third step is the
// creation of a DEM from these gridded points. The fourth step is the
// identification of the original LIDAR points as either BE or OBJ based on
// their relationship to the interpolated
std::vector<PointId> groundIdx;
BOX2D bounds;
view->calculateBounds(bounds);
SpatialReference srs(view->spatialReference());
// Determine the number of rows and columns at the given cell size.
m_numCols = ((bounds.maxx - bounds.minx) / m_cellSize) + 1;
m_numRows = ((bounds.maxy - bounds.miny) / m_cellSize) + 1;
MatrixXd cx(m_numRows, m_numCols);
MatrixXd cy(m_numRows, m_numCols);
for (auto c = 0; c < m_numCols; ++c)
{
for (auto r = 0; r < m_numRows; ++r)
{
cx(r, c) = bounds.minx + (c + 0.5) * m_cellSize;
cy(r, c) = bounds.miny + (r + 0.5) * m_cellSize;
}
}
// STEP 1:
// As with many other ground filtering algorithms, the first step is
// generation of ZImin from the cell size parameter and the extent of the
// data. The two vectors corresponding to [min:cellSize:max] for each
// coordinate – xi and yi – may be supplied by the user or may be easily and
// automatically calculated from the data. Without supplied ranges, the SMRF
// algorithm creates a raster from the ceiling of the minimum to the floor
// of the maximum values for each of the (x,y) dimensions. If the supplied
// cell size parameter is not an integer, the same general rule applies to
// values evenly divisible by the cell size. For example, if cell size is
// equal to 0.5 m, and the x values range from 52345.6 to 52545.4, the range
// would be [52346 52545].
// The minimum surface grid ZImin defined by vectors (xi,yi) is filled with
// the nearest, lowest elevation from the original point cloud (x,y,z)
// values, provided that the distance to the nearest point does not exceed
// the supplied cell size parameter. This provision means that some grid
// points of ZImin will go unfilled. To fill these values, we rely on
// computationally inexpensive image inpainting techniques. Image inpainting
// involves the replacement of the empty cells in an image (or matrix) with
// values calculated from other nearby values. It is a type of interpolation
// technique derived from artistic replacement of damaged portions of
// photographs and paintings, where preservation of texture is an important
// concern (Bertalmio et al., 2000). When empty values are spread through
// the image, and the ratio of filled to empty pixels is quite high, most
// methods of inpainting will produce satisfactory results. In an evaluation
// of inpainting methods on ground identification from the final terrain
// model, we found that Laplacian techniques produced error rates nearly
// three times higher than either an average of the eight nearest neighbors
// or D’Errico’s spring-metaphor inpainting technique (D’Errico, 2004). The
// spring-metaphor technique imagines springs connecting each cell with its
// eight adjacent neighbors, where the inpainted value corresponds to the
// lowest energy state of the set, and where the entire (sparse) set of
// linear equations is solved using partial differential equations. Both of
// these latter techniques were nearly the same with regards to total error,
// with the spring technique performing slightly better than the k-nearest
// neighbor (KNN) approach.
MatrixXd ZImin = eigen::createMinMatrix(*view.get(), m_numRows, m_numCols,
m_cellSize, bounds);
// MatrixXd ZImin_painted = inpaintKnn(cx, cy, ZImin);
// MatrixXd ZImin_painted = TPS(cx, cy, ZImin);
MatrixXd ZImin_painted = expandingTPS(cx, cy, ZImin);
if (!m_outDir.empty())
{
std::string filename = FileUtils::toAbsolutePath("zimin.tif", m_outDir);
eigen::writeMatrix(ZImin, filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("zimin_painted.tif", m_outDir);
eigen::writeMatrix(ZImin_painted, filename, "GTiff", m_cellSize, bounds, srs);
}
ZImin = ZImin_painted;
// STEP 2:
// The second stage of the ground identification algorithm involves the
// application of a progressive morphological filter to the minimum surface
// grid (ZImin). At the first iteration, the filter applies an image opening
// operation to the minimum surface. An opening operation consists of an
// application of an erosion filter followed by a dilation filter. The
// erosion acts to snap relative high values to relative lows, where a
// supplied window radius and shape (or structuring element) defines the
// search neighborhood. The dilation uses the same window radius and
// structuring element, acting to outwardly expand relative highs. Fig. 2
// illustrates an opening operation on a cross section of a transect from
// Sample 1–1 in the ISPRS LIDAR reference dataset (Sithole and Vosselman,
// 2003), following Zhang et al. (2003).
// paper has low point happening later, i guess it doesn't matter too much, this is where he does it in matlab code
MatrixXi Low = progressiveFilter(-ZImin, m_cellSize, 5.0, 1.0);
// matlab code has net cutting occurring here
MatrixXd ZInet = ZImin;
MatrixXi isNetCell = MatrixXi::Zero(m_numRows, m_numCols);
if (m_cutNet > 0.0)
{
MatrixXd bigOpen = eigen::matrixOpen(ZImin, 2*std::ceil(m_cutNet / m_cellSize));
for (auto c = 0; c < m_numCols; c += std::ceil(m_cutNet/m_cellSize))
{
for (auto r = 0; r < m_numRows; ++r)
{
isNetCell(r, c) = 1;
}
}
for (auto c = 0; c < m_numCols; ++c)
{
for (auto r = 0; r < m_numRows; r += std::ceil(m_cutNet/m_cellSize))
{
isNetCell(r, c) = 1;
}
}
for (auto c = 0; c < m_numCols; ++c)
{
for (auto r = 0; r < m_numRows; ++r)
{
if (isNetCell(r, c)==1)
ZInet(r, c) = bigOpen(r, c);
}
}
}
// and finally object detection
MatrixXi Obj = progressiveFilter(ZInet, m_cellSize, m_percentSlope, m_maxWindow);
// STEP 3:
// The end result of the iteration process described above is a binary grid
// where each cell is classified as being either bare earth (BE) or object
// (OBJ). The algorithm then applies this mask to the starting minimum
// surface to eliminate nonground cells. These cells are then inpainted
// according to the same process described previously, producing a
// provisional DEM (ZIpro).
// we currently aren't checking for net cells or empty cells (haven't i already marked empty cells as NaNs?)
MatrixXd ZIpro = ZImin;
for (int i = 0; i < Obj.size(); ++i)
{
if (Obj(i) == 1 || Low(i) == 1 || isNetCell(i) == 1)
ZIpro(i) = std::numeric_limits<double>::quiet_NaN();
}
// MatrixXd ZIpro_painted = inpaintKnn(cx, cy, ZIpro);
// MatrixXd ZIpro_painted = TPS(cx, cy, ZIpro);
MatrixXd ZIpro_painted = expandingTPS(cx, cy, ZIpro);
if (!m_outDir.empty())
{
std::string filename = FileUtils::toAbsolutePath("zilow.tif", m_outDir);
eigen::writeMatrix(Low.cast<double>(), filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("zinet.tif", m_outDir);
eigen::writeMatrix(ZInet, filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("ziobj.tif", m_outDir);
eigen::writeMatrix(Obj.cast<double>(), filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("zipro.tif", m_outDir);
eigen::writeMatrix(ZIpro, filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("zipro_painted.tif", m_outDir);
eigen::writeMatrix(ZIpro_painted, filename, "GTiff", m_cellSize, bounds, srs);
}
ZIpro = ZIpro_painted;
// STEP 4:
// The final step of the algorithm is the identification of ground/object
// LIDAR points. This is accomplished by measuring the vertical distance
// between each LIDAR point and the provisional DEM, and applying a
// threshold calculation. While many authors use a single value for the
// elevation threshold, we suggest that a second parameter be used to
// increase the threshold on steep slopes, transforming the threshold to a
// slope-dependent value. The total permissible distance is then equal to a
// fixed elevation threshold plus the scaling value multiplied by the slope
// of the DEM at each LIDAR point. The rationale behind this approach is
// that small horizontal and vertical displacements yield larger errors on
// steep slopes, and as a result the BE/OBJ threshold distance should be
// more per- missive at these points.
// The calculation requires that both elevation and slope are interpolated
// from the provisional DEM. There are any number of interpolation
// techniques that might be used, and even nearest neighbor approaches work
// quite well, so long as the cell size of the DEM nearly corresponds to the
// resolution of the LIDAR data. A comparison of how well these different
// methods of interpolation perform is given in the next section. Based on
// these results, we find that a splined cubic interpolation provides the
// best results.
// It is common in LIDAR point clouds to have a small number of outliers
// which may be either above or below the terrain surface. While
// above-ground outliers (e.g., a random return from a bird in flight) are
// filtered during the normal algorithm routine, the below-ground outliers
// (e.g., those caused by a reflection) require a separate approach. Early
// in the routine and along a separate processing fork, the minimum surface
// is checked for low outliers by inverting the point cloud in the z-axis
// and applying the filter with parameters (slope = 500%, maxWindowSize =
// 1). The resulting mask is used to flag low outlier cells as OBJ before
// the inpainting of the provisional DEM. This outlier identification
// methodology is functionally the same as that of Zhang et al. (2003).
// The provisional DEM (ZIpro), created by removing OBJ cells from the
// original minimum surface (ZImin) and then inpainting, tends to be less
// smooth than one might wish, especially when the surfaces are to be used
// to create visual products like immersive geographic virtual environments.
// As a result, it is often worthwhile to reinter- polate a final DEM from
// the identified ground points of the original LIDAR data (ZIfin). Surfaces
// created from these data tend to be smoother and more visually satisfying
// than those derived from the provisional DEM.
// Very large (>40m in length) buildings can sometimes prove troublesome to
// remove on highly differentiated terrain. To accommodate the removal of
// such objects, we implemented a feature in the published SMRF algorithm
// which is helpful in removing such features. We accomplish this by
// introducing into the initial minimum surface a ‘‘net’’ of minimum values
// at a spacing equal to the maximum window diameter, where these minimum
// values are found by applying a morphological open operation with a disk
// shaped structuring element of radius (2?wkmax). Since only one example in
// this dataset had features this large (Sample 4–2, a trainyard) we did not
// include this portion of the algorithm in the formal testing procedure,
// though we provide a brief analysis of the effect of using this net filter
// in the next section.
MatrixXd scaled = ZIpro / m_cellSize;
MatrixXd gx = eigen::gradX(scaled);
MatrixXd gy = eigen::gradY(scaled);
MatrixXd gsurfs = (gx.cwiseProduct(gx) + gy.cwiseProduct(gy)).cwiseSqrt();
// MatrixXd gsurfs_painted = inpaintKnn(cx, cy, gsurfs);
// MatrixXd gsurfs_painted = TPS(cx, cy, gsurfs);
MatrixXd gsurfs_painted = expandingTPS(cx, cy, gsurfs);
if (!m_outDir.empty())
{
std::string filename = FileUtils::toAbsolutePath("gx.tif", m_outDir);
eigen::writeMatrix(gx, filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("gy.tif", m_outDir);
eigen::writeMatrix(gy, filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("gsurfs.tif", m_outDir);
eigen::writeMatrix(gsurfs, filename, "GTiff", m_cellSize, bounds, srs);
filename = FileUtils::toAbsolutePath("gsurfs_painted.tif", m_outDir);
eigen::writeMatrix(gsurfs_painted, filename, "GTiff", m_cellSize, bounds, srs);
}
gsurfs = gsurfs_painted;
MatrixXd thresh = (m_threshold + m_scalar * gsurfs.array()).matrix();
if (!m_outDir.empty())
{
std::string filename = FileUtils::toAbsolutePath("thresh.tif", m_outDir);
eigen::writeMatrix(thresh, filename, "GTiff", m_cellSize, bounds, srs);
}
for (PointId i = 0; i < view->size(); ++i)
{
using namespace Dimension;
double x = view->getFieldAs<double>(Id::X, i);
double y = view->getFieldAs<double>(Id::Y, i);
double z = view->getFieldAs<double>(Id::Z, i);
int c = Utils::clamp(static_cast<int>(floor(x - bounds.minx) / m_cellSize), 0, m_numCols-1);
int r = Utils::clamp(static_cast<int>(floor(y - bounds.miny) / m_cellSize), 0, m_numRows-1);
// author uses spline interpolation to get value from ZIpro and gsurfs
if (std::isnan(ZIpro(r, c)))
continue;
// not sure i should just brush this under the rug...
if (std::isnan(gsurfs(r, c)))
continue;
double ez = ZIpro(r, c);
// double ez = interp2(r, c, cx, cy, ZIpro);
// double si = gsurfs(r, c);
// double si = interp2(r, c, cx, cy, gsurfs);
// double reqVal = m_threshold + 1.2 * si;
if (std::abs(ez - z) > thresh(r, c))
continue;
// if (std::abs(ZIpro(r, c) - z) > m_threshold)
// continue;
groundIdx.push_back(i);
}
return groundIdx;
}
TIndexKernel::FileInfo TIndexKernel::getFileInfo(KernelFactory& factory,
const std::string& filename)
{
FileInfo fileInfo;
StageFactory f;
std::string driverName = f.inferReaderDriver(filename);
Stage *s = f.createStage(driverName);
Options ops;
ops.add("filename", filename);
setCommonOptions(ops);
s->setOptions(ops);
applyExtraStageOptionsRecursive(s);
if (m_fastBoundary)
{
QuickInfo qi = s->preview();
std::stringstream polygon;
polygon << "POLYGON ((";
polygon << qi.m_bounds.minx << " " << qi.m_bounds.miny;
polygon << ", " << qi.m_bounds.maxx << " " << qi.m_bounds.miny;
polygon << ", " << qi.m_bounds.maxx << " " << qi.m_bounds.maxy;
polygon << ", " << qi.m_bounds.minx << " " << qi.m_bounds.maxy;
polygon << ", " << qi.m_bounds.minx << " " << qi.m_bounds.miny;
polygon << "))";
fileInfo.m_boundary = polygon.str();
if (!qi.m_srs.empty())
fileInfo.m_srs = qi.m_srs.getWKT();
}
else
{
PointTable table;
Stage *hexer = f.createStage("filters.hexbin");
if (! hexer)
{
std::ostringstream oss;
oss << "Unable to create hexer stage to create boundaries. "
<< "Is PDAL_DRIVER_PATH environment variable set?";
throw pdal_error(oss.str());
}
hexer->setInput(*s);
hexer->prepare(table);
PointViewSet set = hexer->execute(table);
MetadataNode m = table.metadata();
m = m.findChild("filters.hexbin:boundary");
fileInfo.m_boundary = m.value();
PointViewPtr v = *set.begin();
if (!v->spatialReference().empty())
fileInfo.m_srs = v->spatialReference().getWKT();
}
FileUtils::fileTimes(filename, &fileInfo.m_ctime, &fileInfo.m_mtime);
fileInfo.m_filename = filename;
return fileInfo;
}