Generate Ground Control Points

Feature Description

A ground control point (GCP) is a control point located at a specific position and on a specific target in an image, with coordinate information in the mapping coordinate system. Due to its high-precision spatial coordinate data, it can be used in processes such as geometric correction of remote sensing imagery, positioning accuracy verification, and spatial registration, to achieve high-precision geographic reference and location tracking of image data.

Parameter Description

Parameter Name Parameter Interpretation Parameter Type
Dataset

Displays the dataset containing the image used to generate ground control points. Not editable.

 DatasetMosaic
Input Image Type Selects the type of image used to generate ground control points. The default is Panchromatic Image, and can also be switched to Multispectral Image, Forward-looking Image, Rear-view Image, or Front View and Rear View Image depending on the situation. ComboBoxImageType
Error Threshold The error threshold for eliminating gross errors in image matching. The value range is [0,40], the default is 5, and the unit is px. During image matching, the least squares method is used to fit the resulting points, and points greater than the error threshold are removed. The larger the threshold, the more tie points are retained, but the probability of retaining incorrect points increases. Double
Point Distribution Mode

Selects the distribution mode for ground control points. Provides two modes: Conventional and Uniform, with the default being Conventional.

  • Conventional: Divides each scene into N*M sub-regions, and then selects n image blocks of 512*512 size from each sub-region to generate control points. The generated ground control points will cover the entire image as much as possible.
  • Uniform: The generated ground control points will be evenly distributed on each scene image. The number of points is less than that of the conventional distribution, but the distribution is more uniform, suitable for situations with large internal distortion in the image.
 PointDistributionMethod
Number of Blocks in Column Direction

Available when Conventional is selected in Point Distribution Mode.

The number of blocks each scene image is divided into in the column direction. The default value is 4.

 Integer
Number of Blocks in Row Direction

Available when Conventional is selected in Point Distribution Mode.

The number of blocks each scene image is divided into in the row direction. The default value is 4.

 Integer
Matching Method

Available when Conventional is selected in Point Distribution Mode.

Provides the following six matching methods, which can be selected based on data characteristics and requirements. Among them, AFHORP and RIFT support multi-modal data matching; CASP and DEEPFT are based on deep learning and require additional configuration of AI models and installation of the CUDA environment; in general, MOTIF, CASP, or DEEPFT are recommended.

  • MOTIF (default): A template matching algorithm for multi-modal images, characterized by the use of lightweight feature descriptors. MOTIF can overcome nonlinear radiometric distortions caused by differences between SAR and optical images.
  • CASP: A novel cascaded matching pipeline that benefits from integrating high-level features, helping to reduce the computational cost of low-level feature extraction. This pipeline decomposes the matching stage into two progressive stages: first, establishing one-to-many correspondences at a coarser scale as cascaded priors. Then, these priors are used for guidance to determine one-to-one matches at the target scale.[1]
  • DEEPFT: A deep learning-based image matching method.
  • SIFT: A method for extracting distinctive invariant features from images, which can be used for reliable matching between objects or scenes under different viewpoints.
  • RIFT: A feature matching algorithm robust to a wide range of nonlinear radiometric distortions. It enhances the stability of feature detection and overcomes the limitations of feature description based on gradient information.
  • AFHORP: A feature matching algorithm for multi-modal images. AFHORP is highly resistant to radiometric distortions and contrast differences in multi-modal images, and performs excellently in solving problems of direction reversal and phase extreme mutation.
 ImageMatchMethod
Intra-Block Maximum

Available when Conventional is selected in Point Distribution Mode.

The maximum number of points retained in each image block during image matching. The value range is [1,2048], and the default is 256.

 Integer
Number of Seed Points

Available when Uniform is selected in Point Distribution Mode.

Sets the number of seed points for matching corresponding points on each scene image. The value range is [64,6400], and the default is 512. When the image texture is poor, it is recommended to increase the number of ground control points to match enough points and improve the subsequent image quality.

 Integer
Seed Point Search

Available when Uniform is selected in Point Distribution Mode.

Sets the method for searching seed points. The default is Grid Center Point.

  • Corner Point: Points with obvious features within the selected region are used as seed points.
  • Grid Center Point: The center point of the grid is used as the seed point. This search method is random.
 SearchSeedPointMethod
Template Size

Available when Uniform is selected in Point Distribution Mode.

Sets the interval size between seed points. The value range is [1,256], the default is 40, and the unit is px. The larger the template, the more reliable the searched points, but the longer the time taken.

 Integer
Search Radius

Available when Uniform is selected in Point Distribution Mode.

Sets the search radius for image matching seed points. The value range is [0,256], the default value is 40, and the unit is px. The larger the search radius, the larger the matching range and the longer the time taken.

 Double
Semantic Culling of Non-Ground Points Not checked by default. When checked, ground control points in cloud areas and building areas can be automatically removed based on AI semantic technology. Boolean
Cloud Area

Available after checking Semantic Culling of Non-Ground Points.

Checked by default, meaning that ground control points within the cloud area will be automatically removed based on the set dataset. If not checked, ground control points in the cloud area will be retained. The dataset must contain the ImageName field, and the name must correspond to the current image to be processed.

 Boolean
Dataset

Displayed after checking Cloud Area, not editable.

If it is a DOM production related workflow, the cloud amount data from Set Image Path is used.

If it is a DSM production related workflow, the cloud amount data from Set Image Path (DSM/DEM) is used.

 DatasetVector
Building Area Available after checking Semantic Culling of Non-Ground Points. Checked by default, meaning building areas will be automatically identified and ground control points in those areas will be removed. If not checked, ground control points in building areas will be retained. Boolean

Output Result

Generates a GroundControlPoint vector point dataset in the Control Point datasource.

Related Topics

Set Image Path

Set Image Path (DSM/DEM)

Generate Tie Points

Block Adjustment

References

[1] Chen, P., Yu, L., Wan, Y., Pei, Y., Liu, X., Yao, Y., ... & Zhang, Y. (2025). CasP: Improving Semi-Dense Feature Matching Pipeline Leveraging Cascaded Correspondence Priors for Guidance. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 28063-28072).