Method of compressing multi-spectral image

A method of compressing a multi-spectral image formed of a plurality of elemental images having different spectral distribution ranges, comprising:

• a step of extracting pairs of elemental images having respective spectral distribution ranges close to each other from said plurality of elemental images;
• a step of averaging the thus extracted pairs of elemental images to calculate average images, provided that an elemental image that does not form a pair is taken to be as such an average image;
• a step of further extracting pairs of inferior average images having respective spectral distribution ranges close to each other;
• a step of further averaging the thus extracted pairs of inferior average images to calculate superior average images, provided that an inferior average image that does not form a pair is taken to be as such a superior average image;
• a step of hierarchically repeating said averaging steps in which the same elemental images and inferior average images are not used to calculate the average images and the superior average images, respectively, thereby finally calculating a superlative average image of all the elemental images;
• a step of creating differential images obtained by taking differences between all the elemental images and the respective average images and between the inferior average images and the respective superior averages;
• a step of calculating compressed differential images by compressing said differential images;
• and
• a step of recording said superlative average image and said compressed differential images so as to take a hierarchical structure with said superlative average image being positioned at a top corresponding to a hierarchical structure constructed of all the elemental images, all the average images and the superlative average image positioned at the top.