|
Abstract:
|
In the era of big data, large datasets are in abundance, especially in the spatio -temporal domain. Some common examples of spatio-temporal data include climate data, census data, satellite-image data, etc. Data compression is required to improve storage and transmission capacity. In this project, we propose an efficient image compression algorithm that exploits the spatial dependency in the data. Our method combines PCA and Fourier transforms, and exploits the inter dependence among with different variables and geostatistical methods for image reconstruction. A key feature of our approach is that the reconstructed images are coupled with a geostatistical model for the compression errors, opening the door for richer statistical analyses of the reconstructed data. The efficiency of the algorithm has been tested on parts of the CMIP 5 datasets.
|
