---
id: engines/cars
about:
  label: CARS
  description: >-
    Open-source photogrammetry pipeline from CNES that builds digital surface
    models, rasters of ground and rooftop height, from two or more satellite
    images of the same place.
  tags:
    - photogrammetry
    - stereo
    - elevation
    - digital surface model
    - satellite
    - 3d
tool:
  form: open-source Python pipeline
  group: 3D cities and mapping
  homepage: 'https://cars.readthedocs.io/'
  licence: Apache-2.0
  cost: free; the satellite images it needs are not supplied
domains:
  - places
open:
  source: open
  data: partly
  api: unknown
  participation: open
  licence: Apache-2.0
  note: >-
    The Pléiades sample of Giza used in the quick start is under France's Etalab
    open licence.
provenance:
  origin: agent
  sources:
    - title: CNES/cars (README)
      url: 'https://github.com/CNES/cars'
    - title: 'CARS, a satellite multi view stereo pipeline — CARS Documentation'
      url: 'https://cars.readthedocs.io/en/stable/'
    - title: Technical Foundations — CARS Documentation
      url: 'https://cars.readthedocs.io/en/stable/technical_foundations/index.html'
    - title: Installation — CARS Documentation
      url: 'https://cars.readthedocs.io/en/stable/installation.html'
    - title: Sensor examples — CARS Documentation
      url: 'https://cars.readthedocs.io/en/stable/sensor_examples/index.html'
    - title: Release 1.3.0 · CNES/cars · GitHub
      url: 'https://github.com/CNES/cars/releases/tag/1.3.0'
  created: '2026-10-21'
  reviewed: '2026-10-21'
  reviewed_by: 'Claude, from the sources listed'
  status: draft
---
Two eyes judge depth from how far a thing shifts between them, and two satellite images taken from different angles do the same for the ground.[^3] From such pairs CARS builds a digital surface model, a raster grid in which each cell holds the height of the surface seen from above.[^2,3] A grid of heights is not a full 3D mesh, so it is often called 2.5D.[^3]

CNES maintains it on GitHub.[^1] It is the core of the 3D image processing chain for the CO3D satellite mission, and the AI4GEO and Evoland projects use it too.[^1,2] It handles large production runs on a computing cluster and also runs on a personal computer.[^2] The current pipeline is described in a 2026 paper in the ISPRS Archives by D. Youssefi and colleagues.[^1] Release 1.3.0 came out in October 2026.[^6]

The work goes in four steps. First the images are resampled so that a point can move only along one line between them, and the search for a match becomes one-dimensional.[^3] Matching then finds each pixel's partner in the other image.[^3] Crossing the two lines of sight places the point in space, with a longitude, latitude and altitude.[^3] Last, every point is projected onto a grid of heights and colours.[^3] Any image GDAL can read will do, as long as it comes with RPC coefficients, a compact formula linking image pixels to ground positions.[^3] Example setups exist for Pléiades, SPOT 6 and 7, Pléiades Neo and Maxar WorldView.[^5]

Matching can fail on moving objects such as vehicles, in shadows, where one viewpoint is blocked, and on surfaces with too little texture.[^3] Installing takes `pip install cars` on Linux or the `cnes/cars` Docker image.[^1,4] The quick start runs on a free Pléiades sample of the pyramids at Giza.[^1]

For a model of a city or a valley, CARS gives the height of the ground and what stands on it from the stereo images the user brings. The code is under the Apache 2.0 licence.[^1]