A process for the accurate reconstruction of pre-filtered and compressed digital aerial images

Abstract

The study of compression and decompression methods is crucial for storage and/or transmission of large numbers of image data which is required for archiving aerial photographs, satellite images and digital ortho-photos. Hence, the proposed work aims to increment the compression ratio (CR) of digital images in general. While emphasis is made on aerial images, the same principle may find applications to other types of raster based images. The process described here involves the application of pre-defined low-pass filters (i.e. kernels) prior to applying standard image compression encoders. Low-pass filters have the effect of increasing the dependence between neighbouring pixels which can be used to improve the CR. However, for this pre-filtering process to be considered as a compression instrument, it should allow for the original image to be accurately restored from its filtered counterpart.The development of the restoration process presented in this study is based on the theory of least squares and assumes the knowledge of the filtered image and the low-pass filter applied to the original image. The process is a variant of a super-resolution algorithm previously described, but its application and adaptation to the filtering and restoration of images, in this case (but not exclusively) aerial imagery, using a number of scales and filter dimensions is the expansion detailed here. An example of the proposed process is detailed in the ensuing sections. The example is also indicative of the degree of accuracy that can be attained upon applying this process to gray-scale images of different entropies and coded in a lossy or lossless mode.