Python | 获取PCD点云数据强度等信息

-拟墨画扇-

已于 2024-04-20 16:58:01 修改

阅读量429

点赞数 3

分类专栏： Python 文章标签： python 开发语言

于 2024-04-20 16:50:58 首次发布

本文链接：https://blog.csdn.net/sun80760/article/details/138002881

版权

Python 专栏收录该内容

30 篇文章 16 订阅

订阅专栏

最近工作需要，需要获取PCD点云数据的强度等信息，给出open3d和pypcd两种方法获取强度信息。读取的PCD数据头格式如下：

VERSION 0.7
FIELDS x y z intensity laserid timeoffset yawangle
SIZE 4 4 4 1 2 8 4
TYPE F F F U U F F
COUNT 1 1 1 1 1 1 1
WIDTH 230400
HEIGHT 1
VIEWPOINT 0 0 0 1 0 0 0
POINTS 230400
DATA binary_compressed

一、使用open3d库

1、安装

pip install open3d==0.15.1

2、获取点云强度

网上给出通过 open3d.io.read_point_cloud() 的方式只能获取点云的 x,y,z 坐标，是无法获取强度等信息的。要想获取强度信息，open3d给出另一种方法，即通过 open3d.t.io.read_point_cloud().point["intensity"] 方式，转换为numpy类型后获取。

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import open3d as o3d

filePath = "pcd_file.pcd"
pcd_type = o3d.t.io.read_point_cloud(filePath)
pcd_positions = o3d.t.io.read_point_cloud(filePath).point["positions"].numpy()
pcd_intensity = o3d.t.io.read_point_cloud(filePath).point["intensity"].numpy()

print(f"type:\n {pcd_type}")
print(f"positions:\n {pcd_positions}")
print(f"intensity:\n {pcd_intensity}")

代码结果如下：

type:
 PointCloud on CPU:0 [230400 points (Float32)] 
  Attributes: 
    yawangle (dtype = Float32, shape = {230400, 1}), 
    intensity (dtype = UInt8, shape = {230400, 1}), 
    laserid (dtype = UInt16, shape = {230400, 1}), 
    timeoffset (dtype = Float64, shape = {230400, 1}).

positions:
 [[ 7.8483152e+00 -8.6269194e-01 -1.8956374e+00]
 [ 6.6055817e+01 -5.3120365e+00 -1.2601746e+00]
 [ 2.4348343e+01 -1.2318233e+00 -1.8688463e+00]
 ...
 [ 2.2996552e+01  8.9388782e-01 -1.8850820e+00]
 [ 3.6149435e-02 -2.0189864e-04  0.0000000e+00]
 [ 5.2160694e+01  5.1382108e+00 -1.5453696e+00]]

intensity:
 [[10]
 [44]
 [20]
 ...
 [ 8]
 [ 1]
 [58]]

参考：Support reading PCD files with intensities · Issue #3341 · isl-org/Open3D · GitHub

二、使用Pypcd库

1、安装

pip install pypcd3==0.1.1

2、获取点云强度

使用 pypcd.PointCloud.from_path() 方法获取，数据保存在属性 pc_data 中。转换为numpy类型后获取。数据按照PCD数据头的顺序排列，切片获取指定列即可。

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import numpy as np
from pypcd import pypcd

filePath = "pcd_file.pcd"
pcd = np.array(pypcd.PointCloud.from_path(filePath).pc_data)
print(pcd)

代码结果如下：

[(7.8483152e+00, -8.6269194e-01, -1.8956374, 10,  3, 1.71318198e+18,  630.)
 (6.6055817e+01, -5.3120365e+00, -1.2601746, 44, 66, 1.71318198e+18,  460.)
 (2.4348343e+01, -1.2318233e+00, -1.8688463, 20, 33, 1.71318198e+18,  290.)
 ...
 (2.2996552e+01,  8.9388782e-01, -1.885082 ,  8, 30, 1.71318198e+18, -223.)
 (3.6149435e-02, -2.0189864e-04,  0.       ,  1, 93, 1.71318198e+18, -393.)
 (5.2160694e+01,  5.1382108e+00, -1.5453696, 58, 60, 1.71318198e+18, -563.)]