Pan sharpening is an image compositing process that uses high-resolution panchromatic (Pan) images to sharpen the corresponding low-resolution multispectral (MS) images. Pan-sharpened MS images are usually color images with high spatial resolution.

For pan-sharpened image of ALOS-3 advanced product, Brovey Conversion will be applied and we will provide high-resolution (80cm resolution) color images by combining Pan images (80cm resolution) and MS images (3.2m resolution).

Pan Sharpening by HSI (HSV) Conversion

The figure below shows the flow of pan sharpening by HSI (HSV) conversion as a typical example.

HSI are capital letters for Hue, Saturation, and Intensity, which indicates the color index. In this figure, intensity is expressed as value and defined as HSV color space.
RGB means for the three primary wavebands: Red, Green, and Blue.
Pan sharpening is a process that converts a color image into an HSV image, replaces the V value of the intensity of the converted HSV image with the DN value of a high-resolution image (Panchromatic), and converts this back into a color image (RGB).

The procedure for creating a pan-sharpened image is as follows.

First, align the low-resolution color image with the high-resolution image. This refers to the process of matching a low-resolution color image to a high-resolution image (Panchromatic) by means of a geometric correction process.
Secondly, convert the low-resolution color image from RGB to HSV color space and convert it to hue, saturation and intensity images.
Then replace the intensity image with a high-resolution image (Panchromatic).
Finally, the three types of images: the replaced intensity image (high-resolution image; Panchromatic), the hue image, and the saturation image are converted from the HSV color space to RGB and converted into a color image (RGB).

By these steps a pan-sharpened image is produced.

Typical Types of Pan Sharpening Process

There are various processing methods for pan sharpening other than HSI conversion.
Here are some typical processing methods.

■ Brovey Conversion
　Brovey algorithm based on spectral modeling is used for data fusion.
■ HSI Conversion (same as HSV method)
　 The color space of hue, saturation and intensity saturation is used for data fusion.
■ Simple Mean Conversion
　The average of the red, green, and blue values and the panchromatic pixel values are used.
■ Gram-Schmidt Spectral Sharpen Method
　The Gram-Schmidt spectral sharpening algorithm is used to sharpen multispectral data.
■ SFIM (Smoothing Filter-based Intensity Modulation)
　By using a ratio between a high-resolution image and its low pass filtered (with a smoothing filter) image, spatial details can be modulated to low-resolution multispectral image.
■ PCA (Principal Component Analysis)
　Transforms multivariate dataset with correlated variables into uncorrelated variables.
■ HPFA (High-Pass Filter Additive)
　Structural and textural details of the high-resolution image are inserted into the low-resolution image.

Details of each processing method can be found here.