pointcloud-extraction 0.0.4

package consisting of the scripts for operating on the point cloud / photogrammetric data

@geospatial-pipelines/data_preparation:

This package provides the tools and utils for the first stage of the operation after fetching data: allowing users to do operations of the current raw dataset of pointcloud (.las, .laz, .pcd) for operations like: - Cropping (2D/3D cropping) - conversion of the coordinates - combination of point clouds - conversion across various formats.

installation and setup:

• Install from the pip directory: pip install pointcloud_extraction_toolkit.
• Or via the source by cloning the whole geospatial-pipeline and then installing the file via the pip install -r requirements.txt.

Various packages:

package_file	remarks
cameras.utils	consist of methods to use the API across colmap / neuralangelo for photogrammetry
pdal.pipeline_generation	scripts to generate the pipeline for PDAL to do necessary transformations
cropping	Script to crop the given portion that you want to fetch.
threed_pointcloud	script that integrates the open3D for 3D data cropping at microlevel

API's :

There are colab tutorials in test/ folder that explain the various usecase, but now try to fetch the

1. Import the cameras package for the photogrammetry pipeline processing to fetch the pointcloud

Important: You need to setup colmap before this package in order to work.

from pointcloud_extraction_toolkit.cameras.utils import ColmapDataParsing
import os
colmap_progressing = ColmapDataParsing(filepath="demo.mp4", output_dir="./demo_output")

colmap_progressing.convert_photo_to_video(downsampling_rate=5)


## also fetch the image metadata for the algorithm / reviewer in order to showcase the details.


## now fetching the image metadata from the given details in order to later on do the required transformation on the specific frame <> pose basis if needed.
files = './output/imgs'

data_info = {}

for filename in os.listdir(files):
    full_path = os.path.join(files, filename)
    imagemetadata = colmap_progressing.get_image_metadata(full_path)
    data_info[filename].append(imagemetadata)

## and finally the colmap transformation

await colmap_processing.colmap_transformation()

## in the results you seem to see some of the outputs are not compatible with the alignment then run the following method to fix and rerun colmap_transformation().
## analyze_colmap_images(self,camera_bin_path, transform_file, camera_depth, coordinates_adjust = ["0", "0", "0", "1"] )


colmap_processing.analyze_colmap_images(camera_bin_path= files + "stereo/camera.bin" , transform_file= files + "transforms.json", camera_depth = "", coordinates_adjust = [] )

2. Tutorial for 3D data processing directly: