3-D volumetric data and remote sensing data are the most encountered multicomponent images. We will present our research in compression of multicomponent data using JPEG 2000. compression of multicomponent data involves many aspects dictated by the user's requirements. We focus on component decorrelation with Principal Component Filter Banks (PCFB). The Principal Component (Karhunen-Loeve) transform renders the best component decorrelation. The advantages of a Principal Component Filter Bank are: less stroge requirements in the header of the compressed data and less computational time.

WHERE: TEC 205