Image fusion for DOM data production under ENVI

Image fusion is an image processing technology that ressamples low-spatial-resolution multi-spectral images or high-spectral data and high-spatial-resolution single-band images to form a sub-high-resolution multi-spectral image remote sensing, the processed images have both high spatial resolution and multi-spectral features. The key to image fusion is accurate registration of the first two images and selection of fusion methods during processing. Only accurate registration of the Two fusion images can produce satisfactory results. The selection of fusion methods depends on the features of the merged image and the purpose of fusion.

ENVI provides the following fusion methods:

L HSV Transformation

L brovey Transformation

The two methods require that the data have a geographical reference or have the same size. The RGB input band must be 8-bit unsigned data or be selected from the enabled color display.

These two methods are similar. The following describes the brovey transformation operation process.

(1) Open the merged two files and display the low-resolution multi-spectral images in the display.

(2) choose Main Menu> transform> image sharpening> color normalized (brovey). In the select input RGB dialog box, there are two options: from the available band list and from the display window, the former requires that the band must be 8 bits without symbols.

(3) Select RGB in the display window and click OK.

(4) In the color normalized (brovey) output panel, select the sampling method and input file path and file name, and click OK to output the result.

For multi-spectral images, ENVI uses the following fusion technologies:

L Gram-Schmidt

L Principal Component (PC) Transformation

L color normalized (CN) Conversion

L PAN sharping

Among the four methods, the Gram-Schmidt method can maintain the consistency of the image spectral information before and after fusion. It is a high-fidelity remote sensing image fusion method; color normalized (CN) the data must have a central wavelength and FWHM, while the pansharping fusion method needs to be enabled in ENVI Zoom, which is suitable for high-resolution images such as QuickBird and Ikonos.

The four operations are similar. The following describes the Gram-Schmidt operation process with relatively many parameters.

(1) Open the merged two files.

(2) choose Main Menu> transform> imagesharpening> Gram-Schmidt spectral sharpening or choose Main Menu> spectral-> Gram-specific dtspectral sharpening.

(3) Select a low-resolution multi-band image in the select Low Spatial resolutionmulti band input file dialog box, and select a high-resolution single-band image in the select high spatial resolution paninput band dialog box.

(4) In the pop-up Gram-Schmidt spectral sharpening output dialog box, you need to select a method to reduce the high-resolution full-color band. There are four methods that have the following meanings:

L avaverage of low resolution multispectral file: uses the average of multi-spectral bands to simulate low-resolution full color bands.

L select input file: select a single-band image of the same size as the multi-spectral data from an external file to simulate a low-resolution full-color band.

L create by sensor type: select a sensor to simulate a low-resolution full-color band. Optional sensors include: Ikonos, IRS1, KOMPSAT-2, landsat7, QuickBird, and SPOT 5. Choose this method, and the fusion image is calibrated by radiation.

L user defined filter function: select a filter function to simulate low-resolution full-color bands. The fused image is the data that has been calibrated by radiation.

Select the average of low resolution multispectral file method.

(5) Select the sampling method, input path, and file name, and click OK to output.

Figure Gram-Schmidt spectral sharpening output dialog box

The following uses the pansharping method to perform Image Fusion Based on the Spot Image Data bit. Follow these steps to implement the following image fusion steps (envi5.0 ):

1) Open the image:

 

Open an image file

2) Select the usage method:

 

Figure Selection Fusion Method

3) operation implementation-simple and effective parameter settings:

 

Figure operation parameter settings

4) view and verify the results

 

Comparison between full-color and multi-spectral original images

 

Comparison between the full-color image and the merged image