【点云NormalEstimation】python-pcl：法向量估计并存储 - CodeAntenna

3D点云特征描述与提取是点云信息处理中最基础也最关键的一部分，点云分割、重采样、配准、曲面重建等处理的大部分算法，都严重依赖特征描述与提取的结果。

1. 估算PointCloud的表面法线（Surface Normals）

有3种方式

1. K近邻计算法向量
2. 半径近邻计算法向量
3. 积分图像进行正态估计（Normal Estimation Using Integral Images）计算法向量

点云表面法线计算的原理，以K近邻为例，搜索每一个点的K近邻，并以这K+1个点估计一个平面出来，穿过这个点并垂直于这个面的线称之为法向量，也称法线。

点云Python-pcl可视化调用的仍然是 PCL的五大依赖库 之一VTK，如果vtk依赖库没有正确安装，点云的可视化将报错；

以上三种方式的法向量计算，并且打印部分发现的值，源代码如下：

import pcl
import time
from laspy.file import File
import numpy as np
import os
import time
import functools


# 日志耗时装饰器
def log_execution_time(func):
    @functools.wraps(func)
    def wrapper(*args, **kwargs):
        start = time.perf_counter()
        res = func(*args, **kwargs)
        end = time.perf_counter()
        print('【%s】 took %.2f s' % (func.__name__, (end - start)))
        return res

    return wrapper


def readPointCloud(path):
    cloud = pcl.PointCloud()
    if (str(path).endswith(".las")):
        f = File(path, mode='r')
        print('[INFO] points：{}'.format(len(f.points)))

        # 构建点云
        inFile = np.vstack((f.x, f.y, f.z)).transpose()
        cloud.from_array(np.array(inFile, dtype=np.float32))
        return cloud
    else:
        cloud = pcl.load(path)
        return cloud
    return cloud


@log_execution_time
def radiusSearchNormalEstimation(cloud):
    ne = cloud.make_NormalEstimation()
    ne.set_RadiusSearch(0.1)
    normals = ne.compute()
    print(normals.size, type(normals), normals[0], type(normals[0]))
    count = 0
    for i in range(0, normals.size):
        if (str(normals[i][0]) == 'nan'):
            continue
        count = count + 1
    print(count)


@log_execution_time
def kSearchNormalEstimation(cloud):
    ne = cloud.make_NormalEstimation()
    tree = cloud.make_kdtree()
    ne.set_SearchMethod(tree)
    ne.set_KSearch(10)
    normals = ne.compute()
    print(normals.size, type(normals), normals[0])
    count = 0
    for i in range(0, normals.size):
        if (str(normals[i][0]) == 'nan'):
            continue
        count = count + 1
    print(count)


@log_execution_time
def integralImageNormalEstimation(cloud):
    normalEstimation = cloud.make_IntegralImageNormalEstimation()
    normalEstimation.set_NormalEstimation_Method_AVERAGE_3D_GRADIENT()
    normalEstimation.set_MaxDepthChange_Factor(0.02)
    normalEstimation.set_NormalSmoothingSize(10.0)
    normals = normalEstimation.compute()
    print(normals.size, type(normals), normals[0])
    # print("[INFO] normalEstimate 耗时:%.2f秒" % (endTime - startTime))
    count = 0
    for i in range(0, normals.size):
        if (str(normals[i][0]) == 'nan'):
            continue
        count = count + 1
    print(count)

# 上级目录同目录的另一个文件夹路径
path = os.path.abspath(os.path.join(os.getcwd(), "../"))
path = path + "/pcds/table_scene_mug_stereo_textured.pcd"
print(path)

cloud = readPointCloud(path)
kSearchNormalEstimation(cloud)
radiusSearchNormalEstimation(cloud)
integralImageNormalEstimation(cloud)