Fusion and vision algorithm of spectral data based on mapping-evaluating-optimizing methods within multi-section

Color vision of spectral data after reducing dimension and redundancy was researched. The traditional mapping method, cutting or compressing data into the range between 0 and 1, can always lead to loss of the fusion image details. A novel fusion and vision algorithm of spectral data was presented, based on the shifting-mapping-evaluating-optimizing way within multi-section. The first 3 principal component values were achieved by principal component transform (PCT) of the spectral data cube, and assigned respectively to white-black, red-green and yellow-blue channels of opponent color space. The values of standard red-green-blue (sRGB) color space were transformed from opponent color space, and the sRGB values were divided into several sections, moved to the range of 0 to 1 and mapped to 8bit RGB digital code values. The single item evaluations of standard deviation, entropy, average gradient were calculated from 8bit RGB. The comprehensive evaluation values were got from all of the single item evaluation after finishing the process of moving, mapping and single item evaluation. The fusion image was mapped at the section with maximum comprehensive evaluation. The results show that the fusion image can ensure the images energy, information and definition, which is useful for manual distinguish and judge rapidly.