open3d显示RGBD点云(rope3d数据集)

VIP文章流浪德意志

已于 2022-10-02 17:02:27 修改

阅读量573

点赞数

分类专栏： LiDAR目标检测深度视觉目标检测 Python 文章标签： python 人工智能开发语言

于 2022-10-02 17:00:42 首次发布

本文链接：https://blog.csdn.net/li4692625/article/details/127144333

版权

rope3d中同时有深度图,因此可以创建rgbd图像点云

借助open3d库

需要确定相机内参和外参

注意统一坐标系例如都显示在world坐标系

#! /usr/bin/env python3
# -*-coding:utf-8-*-
#for reading pcd and json from task3 result files
import pcl
import open3d as o3d
from pandaset import DataSet, geometry
import numpy as np
from pyquaternion import Quaternion
import pandas
from pathlib import Path
import json
from math import tanh
import os
from show_2d3d_box import read_kitti_ext, load_detect_data, Data, read_kitti_cal 
import math

if __name__ == "__main__":
    folder="/Rope3D/validation"
    extrinsics_dir = os.path.join(folder, 'extrinsics')
    name="148709_sj8fas2e152d20211124air_420_1637216131_1637218737_1_obstacle"
    extrinsic_file = os.path.join(extrinsics_dir, '%s.yaml' % (name))
    world2camera = read_kitti_ext(extrinsic_file).reshape((4, 4))
    world2camera=np.matrix(world2camera)
    camera2worl