Compression speeds image transmission
NASA`s planned Earth Observing System (EOS) will generate immense quantities of image data. According to Andrew Watson, a senior scientist for vision research at NASA Ames Research (Moffett Field, CA), EOS is expected to generate in excess of 1 Tbyte of image data per day.
To handle such large amounts of data, NASA is experimenting with various types of image-compression techniques. At NASA Ames, Watson and his colleage Albert Ahumada have developed two techniques called DCTune algorithms that optimize image-compression techniques for human observation.
The first, developed jointly with IBM (Yorktown Heights, NY), is based on psychophysical measurements and computes a discrete cosine transform (DCT) quantization matrix for a combination of viewing distances, display resolution, and brightness. The second technique optimizes the DCT quantization matrix of each individual image, based on its contents. This is accomplished by modeling how sensitive the human visual system is to compression artifacts.
"Together," says Watson, "these techniques will allow optimization of image compression in NASA imaging systems." Andrew Watson can be reached at firstname.lastname@example.org. nasa.gov.