Skip to content

Latest commit

 

History

History
96 lines (70 loc) · 4.64 KB

README.md

File metadata and controls

96 lines (70 loc) · 4.64 KB
Raw

NYC - DoiTT

Introduction

This repository contains the code needed to build the Docker image geosolutionsit/nyc-doitt that provides a pre-configured environment with GDAL and Python libraries for processing raster geospatial images.

Prerequisites

Usage

The container is designed to run an operation on a single batch of data, the latter identified by a name.

docker run --rm -t \
    -v /absolute/path/to/your/host/data:/usr/src/app/input \
    -v /absolute/path/to/your/host/output:/usr/src/app/output \
    -e OPERATION="deflate" \
    -e NAME="manhattan" \
    geosolutionsit/nyc-doitt:1.0.0
  • -v (Bind Mounts, /absolute/path...): Map your local input and output directories to the corresponding directories in the container.
  • -e OPERATION: Specify the operation to perform: "deflate" or "compression". Defaults to "deflate".
  • -e NAME: Provide a name for your output files. Defaults to "output" if not specified.

Note: The --rm option is included to automatically remove the container after the processing completes, helping to keep your Docker environment tidy. The -t option enables interactive mode, which can be helpful for viewing real-time logs and troubleshooting issues during execution. Feel free to adjust these options or the entire container run command to suit your specific needs and preferences.

JPEG compression (lossy)

The "compression" operation uses a Shell script (scripts/lossy_comp.sh) that leverages the native GDAL implementation. This script converts input JP2 files into TIFF format, employing JPEG compression with a lossless binary mask. It processes all .jp2 files within the specified input folder and generates a single output TIFF file, named according to the NAME variable, in the designated output folder.

To ensure proper execution within a Docker container, the local input and output paths should be bound to the container paths usr/src/app/input and usr/src/app/output, respectively. This volume mapping still applies for deployment in cloud environments such as Kubernetes, where volumes for the input and output should be defined for the deployed container.

Example

Let's assume the following paths for the processing of Brooklyn's aerial images:

  • Input folder path: /home/user/Desktop/JP2000
  • Output folder path: /home/user/Desktop/tiff_output

To process the images, the command would be this:

docker run --rm \
    -v /home/user/Desktop/JP2000:/usr/src/app/input \
    -v /home/user/Desktop/tiff_output:/usr/src/app/output \
    -e OPERATION="compression" \
    -e NAME="brooklyn" \
    geosolutionsit/nyc-doitt:1.0.0

After the execution completes, the output is the single TIFF file /home/user/Desktop/tiff_output/brooklyn.tif:

/home/user/Desktop/tiff_output
├── brooklyn.tif
[...]

COG with deflate compression (lossless)

The "deflate" operation utilizes the Python wrapper for the GDAL library, with the processing code located in scripts/cog_deflate.py. As input, the script expects the path to a folder containing all .jp2 files. The processed output chunks are then saved within the designated output folder, in a directory named according to the NAME variable.

The same considerations regarding the "compression" operation also apply in this case: the input and output paths should be bound to the container paths usr/src/app/input and usr/src/app/output, respectively. This should be done also for deployments in cloud environments, using volumes.

Example

Let's assume this context for the processing of Manhattan's aerial images:

  • Input folder path: /home/user/Desktop/JP2000
  • Output folder path: /home/user/Desktop/processing_output

The command would be the following:

docker run --rm -t \
    -v /home/user/Desktop/JP2000:/usr/src/app/input \
    -v /home/user/Desktop/processing_output:/usr/src/app/output \
    -e OPERATION="deflate" \
    -e NAME="manhattan" \
    geosolutionsit/nyc-doitt:1.0.0

After the execution completes, the output chunks are going to be placed in a folder named "manhattan", inside the output directory /home/user/Desktop/processing_output/manhattan:

/home/user/Desktop/processing_output
├── manhattan
│   ├── chunk_0_0.tif
│   ├── chunk_0_131072.tif
│   ├── chunk_0_65536.tif
│   ├── chunk_65536_0.tif
│   [...]
[...]

Additional information