Chebyshev Data Compression
Both lossy and lossless data compression methods have come into wide use for a variety of data compression needs. Lossless compression algorithms typically achieve reductions in data volume by a factor of 1.5 to 2, essentially without loss of information. Lossy algorithms, on the other hand, achieve compression factors much larger than 2, but with loss of information.
To date, lossy data compression algorithms have found acceptance in applications where the need to achieve high compression ratios outweighs the information loss, but have not found acceptance in technical applications where preservation of absolute data values are important. Typically, these applications involve time series data (such as measured by a magnetometer or a particle sensor) or spectral data (such as the readout of a spectrometer, which also appears as a time series to the data system). To be useful for time series data, a lossy compression algorithm needs to be optimized to preserve absolute data values to the extent possible within the available bandwidth. Researchers in the JHU/APL Space Department have created a new technology for lossy data compression whose purpose is to make data files as small as possible to facilitate data transmission and/or storage while at the same time preserving the necessary information in the file. Versions of the algorithm have been developed in one dimension for time series data, in two dimensions for image data, and in three dimensions for image cube data. The algorithm preserves sufficient fidelity of the original data for scientific applications while achieving high compression ratios. The algorithm has the property of requiring few computations to compress data, and it is therefore well suited to applications requiring real-time data compression in extreme terrestrial or extra-terrestrial environments. The algorithm has been implemented in a space-qualified processor and will be implemented in a digital device suitable for space flight. The digital device implementation will allow integration within sensors like imagers, spectrometers, magnetometers, etc., which can then directly produce compressed data output in real time. This solution will eliminate the need for external data storage, retrieval, and then compression in a separate processor.CONTACT:
Mr. M. T. Hickman