Input Images


Images are the main input data to SURE, next to the corresponding orientation file(s) and (optionally) LiDAR data. There are two ways in which SURE uses the Input Images to generate the output products:

  • Geometric reconstruction of 3D surfaces

  • Generation of Colored Point Clouds, True Orthos and Textured Meshes

Supported Camera Model

SURE can process Digital Frame Images based on the Pinhole Camera Model and supports various lens distortion models. See also additional information about the projection geometry.

Supported Image Formats

  • The supported image formats are tif, png and jpg

  • tif and png can be supplied in 8 bit or 16 bit; jpg only supports 8 bit

  • Image files can contain up to 4 channels, typically Red, Green, Blue (+ optionally Infrared)

Image Quality

The quality of the Image content is crucial for obtaining optimal results. High quality texture implies the possibility to distinguish details, down to the pixel local context. In addition to this, having sufficient coverage and redundancy is also important for obtaining a complete and accurate surface reconstruction. More information about this is available at Image Quality and Image Acquisition and Overlap.

SURE supports the use of image files with LZW, Deflate or JPEG compression. Avoid using strongly compressed images. These can increase the risk of compression artefacts that reduce the quality of reconstruction.

If enabled, SURE applies corrections in the color space, affecting exclusively the texture of output products. See more information about this in the Image Processing Features article.


By default, SURE will use the specified Input Images for geometric reconstruction of 3D surfaces and for applying texture on Point Clouds, True Orthos and Textured Meshes.

The user also has the possibility to specify separate Matching Images which will be used exclusively for the surface reconstruction (e.g. for Dense Matching). In this situation, it is still necessary to specify the Input Images which will be used for the texturing. There must be a direct correspondence (pixel to pixel) between the Matching Images and the Texture Images. Consequently, the file names should correspond to the same orientation identifier.


The Input Images are specified during the Project Configuration wizard.


 -i, --images <path>              Set path to input images (optional for Inpho prj
                                   orientations with valid image paths)
 --matching-images <path>         Set path to separate input images for matching

Image Preparation

The Image Preparation step makes sure that the input imagery can be used optimally during subsequent processing steps.

During Image Preparation, the following operations can potentially take place:

  • removing distortion → based on the distortion model parameters stored in the orientation data

  • writing of Images in Tiled Tiff format → 256 x 256 pixels blocks. This ensures faster access to image content, accelerating various processing steps.

  • generating Image Pyramids → stored internally in the image files, accelerating various processing steps.

The results of the Image Preparation step are the prepared images stored inside Internal/Images folder.

If none of the operations in the Image Preparation step are necessary, because the conditions are already met, the Image Preparation step is skipped entirely, saving processing time and disk storage.

To skip the Image Preparation step, the Input Images must meet the following conditions:

  • undistorted → The information that the images don’t contain any distortion is read from the orientation data.

  • Tiled Tiff → E.g. 256 x 256 pixels blocks

  • contain pyramids that have been generated by compatible software (Match AT / SURE). The pyramids may be either internal or stored in separate files (*.pyr).

SURE will automatically check and decide for each image, whether preparation is necessary.

Best Practice

  • Use Input Images with high texture quality.

  • If available, specify 16 bit Imagery. In particular as Matching Images, they will tend to deliver more accurate 3D reconstruction results.

  • Avoid using strongly compressed Images.

  • Whenever possible, use Input Images that don't require Preparation. This will reduce processing time and disk storage.

    • In this scenario, it is highly recommended to store the Input Images on fast access devices (e.g. RAID SSD), to achieve the best performance.

  • See more information at Production with SURE - Best Practices

Further Reading

Lens Distortion Models

Coordinate Systems

Image Quality

Image Acquisition and Overlap

Image Processing Features

SURE Project Configuration

Production with SURE - Best Practices