ROI-Based On-Board Compression for Hyperspectral Remote Sensing Images on GPU.

摘要: In recent years, hyperspectral sensors for Earth remote sensing have become very popular. Such systems are able to provide the user with images having both spectral and spatial information. The current spaceborne capture large areas increased resolution. For this reason, volume of acquired data needs be reduced on board in order avoid a low orbital duty cycle due limited storage space. Recently, literature has focused attention efficient ways on-board compression. This topic is challenging task difficult environment (outer space) time, power computing resources. Often, hardware properties Graphic Processing Units (GPU) been adopted reduce processing time using parallel computing. work proposes framework operation GPU, NVIDIA’s CUDA (Compute Unified Device Architecture) architecture. algorithm aims at performing compression target’s related strategy. detail, main operations are: automatic recognition land cover types or detection events near real regions interest (this choice) an unsupervised classifier; specific space-variant different bit rates including Principal Component Analysis (PCA), wavelet arithmetic coding; management Ground Station. Experiments provided dataset taken from AVIRIS (Airborne Visible/Infrared Imaging Spectrometer) airborne sensor harbor area.