PCL point cloud library, feature description extraction integral image (Integral Image)

When I was learning PCL, I encountered an example. For the ordered point cloud , the integral image (Integral Image) operation can be performed . The point cloud library provides 4 implementation methods:

        COVARIANCE_MATRIX, //局部协方差矩阵，建立9个积分图来计算法线
        AVERAGE_3D_GRADIENT,//创建6个积分图来计算水平垂直方向平滑后的三维梯度
        AVERAGE_DEPTH_CHANGE,//创建1个单一积分图，并从深度变化计算法线
        SIMPLE_3D_GRADIENT //

 

 You can compare the effects of the first three and consolidate to achieve multi-window display.

code snippet:

 #include <pcl/io/io.h>
 #include <pcl/io/pcd_io.h>
 #include <pcl/features/integral_image_normal.h>//Integral images 
 #include <pcl/visualization/cloud_viewer.h>
     int
     main ()
     {
             //set up cloud
             pcl::PointCloud<pcl::PointXYZ>::Ptr cloud (new pcl::PointCloud<pcl::PointXYZ>);
             pcl::io::loadPCDFile ("table_scene_mug_stereo_textured.pcd", *cloud);
             //set up normal
             pcl::PointCloud<pcl::Normal>::Ptr normals_1 (new pcl::PointCloud<pcl::Normal>);
             pcl::PointCloud<pcl::Normal>::Ptr normals_2 (new pcl::PointCloud<pcl::Normal>);
             pcl::PointCloud<pcl::Normal>::Ptr normals_3 (new pcl::PointCloud<pcl::Normal>);
             std::vector<pcl::PointCloud<pcl::Normal>::Ptr> NormalPt {normals_1,normals_2,normals_3};

             pcl::IntegralImageNormalEstimation<pcl::PointXYZ,pcl::Normal> ne_1;
             pcl::IntegralImageNormalEstimation<pcl::PointXYZ,pcl::Normal> ne_2;
             pcl::IntegralImageNormalEstimation<pcl::PointXYZ,pcl::Normal> ne_3;

               ne_1.setInputCloud(cloud);
               ne_2.setInputCloud(cloud);
               ne_3.setInputCloud(cloud);
               ne_1.setNormalEstimationMethod (ne_1.AVERAGE_3D_GRADIENT);
               ne_2.setNormalEstimationMethod (ne_2.COVARIANCE_MATRIX);
               ne_3.setNormalEstimationMethod (ne_3.AVERAGE_DEPTH_CHANGE);
               ne_1.setMaxDepthChangeFactor(0.02f);
               ne_1.setNormalSmoothingSize(10.0f);
               ne_1.compute(*NormalPt[0]);
               ne_2.setMaxDepthChangeFactor(0.02f);
               ne_2.setNormalSmoothingSize(10.0f);
               ne_2.compute(*NormalPt[1]);
               ne_3.setMaxDepthChangeFactor(0.02f);
               ne_3.setNormalSmoothingSize(10.0f);
               ne_3.compute(*NormalPt[2]);

              // std::vector<pcl::IntegralImageNormalEstimation<pcl::PointXYZ, pcl::Normal>> ne {ne_1,ne_2,ne_3};

            //visualization
             pcl::visualization::PCLVisualizer viewer("PCL Viewer");
             viewer.initCameraParameters();
             int v1 (0);
             int v2 (0);
             int v3 (0);
            //分割窗口
             viewer.createViewPort(0,0,0.3333,1,v1);
             viewer.setBackgroundColor (0.0, 0.0, 0.5,v1);
             viewer.addText("AVERAGE_3D_GRADIENT",10,10,"v1",v1);
             viewer.createViewPort(0.3333,0.0,0.6666,1,v2);
             viewer.setBackgroundColor (0.3, 0.3, 0.5,v2);
             viewer.addText("COVARIANCE_MATRIX",10,10,"v2",v2);
             viewer.createViewPort(0.66666,0,1,1,v3);
             viewer.setBackgroundColor (0.6, 0.3, 0.1,v3);
             viewer.addText("AVERAGE_DEPTH_CHANGE",10,10,"v3",v3);
             //viewer.addPointCloud<pcl::PointXYZ>(cloud,"cloud1",v1);
             //viewer.addPointCloud<pcl::PointXYZ>(cloud,"cloud2",v2);
             //viewer.addPointCloud<pcl::PointXYZ>(cloud,"cloud3",v3);
             viewer.addPointCloudNormals<pcl::PointXYZ,pcl::Normal>(cloud, normals_1,100,0.05f,"normals_1",v1);
             viewer.addPointCloudNormals<pcl::PointXYZ,pcl::Normal>(cloud, normals_2,100,0.05f,"normals_2",v2);
             viewer.addPointCloudNormals<pcl::PointXYZ,pcl::Normal>(cloud, normals_3,100,0.05f,"normals_3",v3);

             while (!viewer.wasStopped ())
             {
               viewer.spinOnce ();
             }
             return 0;
     }

CMakeLists:

cmake_minimum_required(VERSION 2.8 FATAL_ERROR)
project(normal_estimation_using_integral_images)
find_package(PCL REQUIRED)
include_directories(${PCL_INCLUDE_DIRS})
link_directories(${PCL_LIBRARY_DIRS})
add_definitions(${PCL_DEFINITIONS})	
add_executable (normal_estimation_using_integral_images normal_estimation_using_integral_images.cpp)
target_link_libraries (normal_estimation_using_integral_images ${PCL_LIBRARIES})

 

The output is:

 

Summarize:

(1) Key functions:

void pcl::IntegralImageNormalEstimation< PointInT, PointOutT >::setNormalEstimationMethod(NormalEstimationMethod normal_estimation_method)

 (2) Understanding of multi-window contrast display

 function:

void pcl::visualization::PCLVisualizer::createViewPort	(	double	xmin ,
		double	ymin ,
		double	xmax ,
		double	ymax ,
		int &	viewport
	)

 Among them, the maximum and minimum values ​​of x and y are commonly understood as the proportion of the window from xx to xx. In this three comparisons, x is divided into three equal parts. y is still 0~1.

function:

addText has three overloads:

addText() [1/3]

bool pcl::visualization::PCLVisualizer::addText	(	const std::string &	text,
		int	xpos,
		int	ypos ,
		const std::string &	id = "",
		int	viewport = 0
	)

 

 

 

addText() [2/3]

bool pcl::visualization::PCLVisualizer::addText	(	const std::string &	text,
		int	xpos,
		int	ypos ,
		double	r,
		double	g,
		double	b,
		const std::string &	id = "",
		int	viewport = 0
	)

 colored text

 

 with size and color

addText() [3/3]

bool pcl::visualization::PCLVisualizer::addText	(	const std::string &	text,
		int	xpos,
		int	ypos ,
		int	fontsize ,
		double	r,
		double	g,
		double	b,
		const std::string &	id = "",
		int	viewport = 0
	)