public void TestIntegralFile()
{
//copying integral image tutorial from http://pointclouds.org/documentation/tutorials/normal_estimation_using_integral_images.php#normal-estimation-using-integral-images
using (var cloud = new PointCloudOfXYZ())
using (var normals = new PointCloudOfNormal())
using (var ne = new IntegralImageNormalEstimationPointXYZAndNormal())
{
using (var reader = new PCDReader())
if (reader.Read(DataPath("tutorials/table_scene_mug_stereo_textured.pcd"), cloud) < 0)
{
Assert.Fail("could not open pcd file");
}
ne.SetNormalEstimationMethod(IntegralImageNormalEstimation.NormalEstimationMethod.Average3DGradient);
ne.SetMaxDepthChangeFactor(0.02f);
ne.SetNormalSmoothingSize(10f);
ne.SetInputCloud(cloud);
ne.Compute(normals);
Assert.IsTrue(normals.Count == cloud.Count, "normals count did not match cloud count");
}
}
public unsafe void SupervoxelClusteringTutorialTest()
{
var voxelResolution = 0.008f;
var seedResolution = 0.1f;
var colorImportance = 0.2f;
var spatialImportance = 0.4f;
var normalImportance = 1f;
using (var cloud = new PointCloudOfXYZRGBA())
using (var adjacentSupervoxelCenters = new PointCloudOfXYZRGBA())
{
using (var reader = new PCDReader())
//Assert.AreEqual(0, reader.Read(DataPath("tutorials/correspondence_grouping/milk_cartoon_all_small_clorox.pcd"), cloud));
Assert.AreEqual(0, reader.Read(DataPath("tutorials/table_scene_mug_stereo_textured.pcd"), cloud));
var min = new Vector3(float.MaxValue);
var max = new Vector3(float.MinValue);
foreach (var p in cloud.Points)
{
min = Vector3.Min(min, p.V);
max = Vector3.Max(max, p.V);
}
using (var normals = new PointCloudOfNormal(cloud.Width, cloud.Height))
using (var super = new Segmentation.SupervoxelClusteringOfXYZRGBA(voxelResolution, seedResolution))
using (var clusters = new Segmentation.SupervoxelClustersOfXYZRGBA())
{
using (var ne = new IntegralImageNormalEstimationPointXYZAndNormal())
using (var noColor = new PointCloudOfXYZ(cloud.Width, cloud.Height))
{
var count = cloud.Count;
var cptr = cloud.Data;
var nptr = noColor.Data;
for (var i = 0; i < count; i++)
{
(nptr + i)->V = (cptr + i)->V;
}
ne.SetNormalEstimationMethod(IntegralImageNormalEstimation.NormalEstimationMethod.Average3DGradient);
ne.SetMaxDepthChangeFactor(0.02f);
ne.SetNormalSmoothingSize(10f);
ne.SetInputCloud(noColor);
ne.Compute(normals);
}
super.SetInputCloud(cloud);
super.SetNormalCloud(normals);
super.SetColorImportance(colorImportance);
super.SetSpatialImportance(spatialImportance);
super.SetNormalImportance(normalImportance);
super.Extract(clusters);
Assert.IsTrue(clusters.Count > 0);
using (var adjacency = new MultiMapOfuintAnduint())
{
super.GetSupervoxelAdjacency(adjacency);
Assert.AreEqual(350, adjacency.Count);
var i = 0;
foreach (var kvp in adjacency)
{
i++;
}
Assert.AreEqual(350, i);
using (var labelItr = adjacency.Begin())
using (var end = adjacency.End())
{
for (; !labelItr.Equals(end);)
{
var supervoxelLabel = labelItr.Key;
var supervoxel = clusters.At(supervoxelLabel);
foreach (var kvp in adjacency.EqualRange(supervoxelLabel))
{
var neighbor = clusters.At(kvp.Value);
adjacentSupervoxelCenters.Add(neighbor.Centroid);
}
adjacency.UpperBound(supervoxelLabel, labelItr);
}
}
Assert.AreEqual(350, adjacentSupervoxelCenters.Count);
}
}
}
}
