Image Fusion

Feature Description

Image fusion is a technique that composites and transforms multiple images from different times, sensor systems, resolutions, and spectral bands using various methods to generate new imagery. For example, fusing a panchromatic orthophoto with a multispectral orthophoto can yield high-quality imagery that combines high spatial resolution with high spectral resolution. This technology effectively integrates the advantages of various types of images, providing richer information support for subsequent analysis and applications.

SuperMap iDesktopX supports this feature starting from version 11i(2023).

Figure: Multispectral Image Before Fusion Figure: Panchromatic Image Before Fusion Figure: Multispectral Image After Fusion

Function Entrance

Satellite Image Processing Tab->DOM Data Production Group->Image Fusion.

Parameter Description

Parameter Name Parameter Interpretation Parameter Type
Data Type

This parameter is only available when the function is launched from the Satellite Image Processing Tab->DOM Data Production Group->Image Fusion.

Selects the source of input image data, providing two options: Image Files and Mosaic Dataset. The default is Image Files. Use Add, Select All, Select Reverse, Delete, and Settings functions under Image File List to manage image files.

  • Add: Provides four ways to input image files: Add File, Add Folder, Add List, and Add Mosaic Data. When selecting Add Folder, all image files within that folder (including subfolders) are automatically retrieved.
  • Select All: Selects all records in the list.
  • Select Reverse: Inversely selects all records in the list.
  • Delete: Deletes the selected records from the list.
  • Settings: When the input panchromatic and multispectral images are not fully paired, use the Settings button to manage the Pairing Rules to complete pairing for all images.

 
Panchromatic Data Source

This parameter is only available when the function is launched from the Satellite Image Processing Tab->DOM Data Production Group->Image Fusion.

After selecting Mosaic Dataset in Data Type, specifies the data source where the panchromatic image data resides.

DataSource
Panchromatic Data After selecting Mosaic Dataset in Data Type, specifies the mosaic dataset where the panchromatic image data resides. For workflow production, it is display-only and cannot be edited. DatasetMosaic
Multispectral Data Source

This parameter is only available when the function is launched from the Satellite Image Processing Tab->DOM Data Production Group->Image Fusion.

After selecting Mosaic Dataset in Data Type, specifies the data source where the multispectral image data resides.

DataSource
Multispectral Data After selecting Mosaic Dataset in Data Type, specifies the mosaic dataset where the multispectral image data resides. For workflow production, it is display-only and cannot be edited. DatasetMosaic
Fusion Method

Specifies the image fusion method. The default method is Gram-Schmidt.

  • PanSharp: A gray-level relationship-based image fusion method using the least squares principle. It establishes a linear regression model between the multispectral image and the panchromatic image, adjusts the gray values of each band in the multispectral image to match the radiation characteristics of the panchromatic image, thereby reducing color bias after fusion. Suitable for remote sensing applications requiring high spectral accuracy, such as agricultural monitoring and environmental assessment.
  • Gram-Schmidt: An improved method based on Gram-Schmidt orthogonal transformation. It adaptively adjusts the linear relationship between each band of the multispectral image and the panchromatic image to generate a brightness component better suited for practical needs. This method features high spectral fidelity and good spatial detail preservation.
  • CNBROVEY: CNBROVEY (full name Color Normalized Brovey) is an image fusion method based on an improved Brovey transform. By introducing a color normalization strategy, it optimizes the fusion process between multispectral and panchromatic images. It enhances spatial detail while preserving original spectral characteristics as much as possible, suitable for scenarios balancing spectrum and detail.
  • HSV: HSV (full name Hue, Saturation, Value) is an image fusion method based on color space. It converts the multispectral image from RGB color space to HSV space, then replaces or enhances the value (brightness) component using the high spatial resolution of the panchromatic image, thereby improving image detail while preserving hue and saturation. Suitable for RGB band fusion, but performs poorly on non-visible bands (e.g., near-infrared). Only supports 8-bit image data.
  • PCA: PCA (full name Principal Component Analysis) is an image fusion method based on statistical feature extraction. It transforms the multispectral image into principal component space and replaces the first principal component (PC1) with the panchromatic image to achieve effective fusion of spatial detail and spectral information. This method enhances the spatial detail of multispectral images while retaining most spectral information, suitable for remote sensing applications requiring high spatial resolution. However, there is a risk of spectral distortion, and spectral fidelity is average.
  • NND: NND (full name Non-Negative Decomposition) is an image fusion method based on Non-Negative Matrix Factorization (NMF). It jointly decomposes multi-source imagery (e.g., multispectral, panchromatic, infrared) to extract shared features (e.g., spatial structure) and independent features (e.g., spectrum, thermal radiation), and performs fusion based on this. Suitable for fusion between multi-modal data, but computation time is relatively longer.
FusionMethod
Predefined Fusion Weights Check this box to use predefined weights for fusion, but it is only effective for the PanSharp and Gram-Schmidt methods. If unchecked, adaptive weights are used for fusion. Boolean
Sharpening Factor Used to adjust the clarity of the fused image. The default is 1. Double
Resampling Method Sets the resampling method after image fusion. Provides three options: Nearest, Bilinear, and Cubic. The default is Bilinear. For more information on methods, please refer to Resampling Method. ResampleMethod
Export Directory

Sets the storage path for the result image.

If executing a production workflow, uses the Set the Image Path or Add Orthophoto Processing parameter, and this field is not editable.

File
Output Format

Sets the storage type for the result image. Provides four options: Undefined, Tiff/GeoTIFF, Erdas Image, and PCIDSK. The default is Tiff/GeoTIFF. Selecting Undefined means outputting using the image's original storage type.

If executing a production workflow, uses the Output Format parameter from Set the Image Path or Add Orthophoto, and this field is not editable.

FileType
Compression Method

Sets the compression method for the output image. Only available when the output format is Tiff/GeoTIFF. The default is NONE.

If executing a production workflow, uses the Compression Method parameter from Set the Image Path or Add Orthophoto, and this field is not editable.

CompressMethod
Add Results to Map Window

This parameter is only available when the function is launched from the Satellite Image Processing Tab->DOM Data Production Group->Image Fusion.

Checked by default, meaning the result data will be added to the map window for display upon successful function execution.

Boolean
Create an Image Pyramid

Checked by default. Builds pyramids for the image data according to the specified resampling method to improve image zoom display rendering and performance.

If executing a production workflow, uses the Create an Image Pyramid parameter from Set the Image Path or Add Orthophoto, and this parameter is not displayed.

Boolean
Resampling Method

Available after checking Create an Image Pyramid. The default is Average. For more information on methods, please refer to Resampling Method.

If executing a production workflow, uses the Resampling Method parameter from Set the Image Path or Add Orthophoto, and this parameter is not displayed.

PyramidResampleType
Asynchronously Create Statistics

During image browsing, the layer will be stretched based on statistical information. If the image lacks statistical information, it will be displayed using the default stretch method, and 8-bit data will not be stretched for display. Checked by default, meaning statistics will be created for the image in the background without affecting workflow execution. If unchecked, statistics are created while executing this tool, which may consume time.

If executing a production workflow, uses the Asynchronously Create Statistics parameter from Set the Image Path or Add Orthophoto, and this parameter is not displayed.

Boolean

Output Result

  • Outputs image data combining features of panchromatic and multispectral images to the local folder.
  • Automatically creates an image fusion mosaic dataset in the OrthoData data source based on the output image data.

Related Topics

Generate Orthophoto

Multispectral and Panchromatic Registration

True Color Output

Image Uniform Color

Image Mosaic