Remote Sensing Data Compression
The interest in remote sensing images is growing at an enormous pace in the last years. However, transmission and storage of remote sensing images pose a special challenge, and multiple efficient image compression systems have appeared. This chapter contr
- PDF / 2,877,190 Bytes
- 35 Pages / 430 x 660 pts Page_size
- 15 Downloads / 237 Views
1 Introduction Remote sensing is the capability of extracting information about an object without being in physical contact with it [15]. Materials comprising the various objects in a scene reflect, absorb, and emit, electromagnetic radiation depending on their molecular composition and shape. By measuring the radiation over a broad spectral range, the spectrum may be used to identify the materials. In the last years, several sensors have come into play, allowing applications of remote sensing images such as target detection, military surveillance, farming practices assessment, fire detection, weather forecasting,. . . [37]. In addition, modern sensors have improved their acquisition capabilities, so that radiometric resolution, spatial resolution, spectral resolution, and time resolution have increased at least one order of magnitude, enabling better results in the aforementioned applications. Nevertheless, these improved capabilities come also at the price of an increase in the data volume size. Remote sensing sensors are either airborne or spaceborne. For instance, the NASA Jet Propulsion Laboratory (JPL) Airborne Visible/Infrared Imaging Spectrometer [50] records the visible and the near-infrared spectrum of the reflected light, and is capable of producing images of size 2048 rows × 614 columns × 224 bands × 2 bytes per pixel for a total of 563.347.456 bytes (over 537 Megabytes) per flight. Also, the NASA JPL Atmospheric Infrared Sounder [51], which is a standard ultraspectral sounder data, records thousands of bands from the infrared spectrum, produces 240 granules per day, each of size 135 rows × 90 columns × 2107 bands × 2 bytes per pixel, for a total of 12.288.024.000 bytes (over 12 Gigabytes) data daily. Finally, NASA’s project Earth Observing System satellite [49] will generate data at an unprecedented rate, estimated in over a M. Gra˜ na and R.J. Duro (Eds.): Comput. Intel. for Remote Sensing, SCI 133, pp. 27–61, 2008. c Springer-Verlag Berlin Heidelberg 2008 springerlink.com
28
J. Serra-Sagrist` a and F. Aul´ı-Llin` as
Terabyte of data every day. These enormous data volumes ask for the aid of data compression. Data compression may be beneficial to both data storage and to data transmission. Data compression may be broadly divided in three classes: lossless compression refers to the process whereby the decoded image after coding and decoding is exactly the same as the input image, and achieves compression factors of at most 4:1; near-lossless compression refers to the process whereby the decoded image after coding and decoding is perceptually equivalent to the input image, and achieves compression factors of about 30:1; lossy compression refers to the process whereby some information is discarded, leading to much higher compression ratios. The data that may be discarded depends on the final application, so that in some scenarios, lossy-to-lossless compression is desired. Data compression for remote sensing images is effective because, as natural images, they exhibit large spatial correlation. In addition, t
Data Loading...
