Stereo photogrammetry - PIX4Dmatic

What is stereophotogrammetry?

Stereo photogrammetry is a specialized field within photogrammetry that uses the concept of stereoscopy to create three-dimensional representations of objects and environments from two-dimensional photographs. 

What is stereoscopy?

Stereoscopy is a technique that mimics how human eyes perceive depth. It creates the illusion of three-dimensional depth by presenting two slightly different images of the same scene, each captured from a distinct perspective. 

Image Acquisition and Overlap

To create this depth illusion, image overlap is essential. Overlap means that a significant portion of one image also appears in the next. This is similar to how our eyes view almost the same scene from slightly different points.

Image acquisition must be done carefully to ensure this overlap. While various techniques and platforms can be used, in stereo photogrammetry, it's necessary to maintain an overlap of 80% and capture at least two images of the same subject from different positions.

Key Requirements for a Stereo Photogrammetry Project

A stereo photogrammetry project primarily relies on understanding three aspects: the internal orientation of the camera and the external orientation of the images, and undistorted images

Internal orientation refers to the intrinsic characteristics of the camera itself. This includes its focal length, the principal point of symmetry, and the distortion function. This information is determined during a process called camera calibration.

External orientation defines the exact position and angular alignment of each image in three-dimensional space at the moment it was taken. This encompasses its location (X, Y, Z coordinates) and its rotation (often described as roll, pitch, and yaw or omega, phi, kappa).

Undistorted Images

The images used in stereo photogrammetry should ideally be undistorted. This means any lens distortions present in the raw images have been corrected. This correction occurs during the camera calibration process, where the internal orientation parameters are used to adjust each pixel's position. By applying these parameters, distorted images are transformed into undistorted ones, which is crucial for accurate 3D reconstruction.

Original image: The square grid is distorted because of the lens of the camera.	Undistorted image: The square grid is now perfectly aligned.

For professional large format cameras, this is not an issue when using undistorted images since these cameras are calibrated in a laboratory, the internal parameters are precisely measured; consequently, the images produced by these cameras can be utilized directly. 

For drone cameras, images begin to be used, but these parameters must be calculated to undistort them.

Starting with version 1.80 of PIX4Dmatic, it is possible to export undistorted images from a project. Before exporting, the images must first be calibrated through the calibration process. Once calibration is complete, the export option will be available.

The undistorted camera exportation will generate the following files:

• Undistorted images
• CAM file that contains the principal position of symmetry offset X_poff and Y_poff
• SSK camera file with the information related to the camera.

Why Prinpial Point of Symmetry (PPS) instead of Principal Point of Autocollimation (PPA)?

The PPA (Principal Point of Autocollimation) is where the optical axis intersects the image plane.

The PPS (Principal Point of Symmetry), slightly offset, indicating the true center around which lens distortions are most symmetrical and corrected in photogrammetry.