Predictive Lossless Compression of Regions of Interest in Hyperspectral Images With No-Data Regions

ABSTRACT:

This project tends to the issue of proficient prescient lossless pressure on the locales of intrigue (ROIs) in the hyperspectral pictures without any information districts. We propose a two-organize expectation conspire, where a setting closeness based weighted normal forecast is trailed by recursive slightest square separating to decorrelate the hyperspectral pictures for pressure.We at that point propose to apply isolate Golomb-Rice codes for coding the forecast residuals of the full-setting pixels and limit pixels, individually.

To examine the coding additions of this different coding plan, we acquaint a blend geometric model with speak to the residuals related with different mixes of the full-setting pixels and limit pixels. Both data theoretic examination and reproductions on manufactured information affirm the benefit of the different coding plan over the ordinary coding strategy in view of a solitary basic geometric dispersion. We apply the previously mentioned expectation and coding techniques to four freely accessible hyperspectral picture informational collections, achieving critical enhancements more than a few other best in class strategies, including the shape-versatile JPEG 2000 strategy.

BASE PAPER: Predictive Lossless Compression

LEAVE A REPLY

Please enter your comment!
Please enter your name here