COMPRESSION OF ASTRONOMICAL IMAGES BASED ON THE KARHUNEN-LOEVE TRANSFORM

摘要

The study deals with possibility of the Karhunen - Loeve Transform (KLT), also known as PCA (Principal Component Analysis), for astronomical data compression. This approach is based on analysis of statistical properties of these kinds of data. Image data are divided into four groups - dark frame (DF), flat field (FF), light image (LI) and deep sky light image (DSLI). Quality measurement of the influence of loss part of the coder has been based on standard signal functions (mean square error - MSE, peak signal to noise ratio PSNR), subjective quality measurement according to ITU standard (International Telecommunication Union) and the efficiency astronomical image processing. The astronomical (scientific) image processing include star position measurement (astrometry), brightness measurement (photometry), estimation of deformation the point spread function of the detection (PSF) and etc. The total amount of 20% KLT spectral components has been determined as a sufficient number for high precision measurement. The course of covariance matrix eigenvalues (and eigenimages also) analysis has been carried out. Real images from BOOTES experiment (Burst Observer of Optical Transient Exploring System) have been used as testing signals.