SKAIPACK

The CCSDS-123.0-B-2 implementation provides lossless and near lossless compression specifically designed for multispectral and hyperspectral images. This standard employs a prediction-based algorithm followed by entropy coding, achieving compression ratios of 2:1 to 4:1 while maintaining perfect data fidelity. The FPGA IP core implementation features a streamlined architecture supporting Band Interleaved by Pixel (BIP) sample ordering with Sample-Adaptive entropy coding.

SKAIPACK JPEG2000 delivers high-performance, standards-based image compression for Earth observation, supporting both lossless and lossy modes under ISO/IEC 15444-1. It enables progressive transmission and scalable decoding using DWT and EBCOT, with compression ratios up to 2.5:1 (lossless) and 70:1 (lossy). Advanced Region of Interest (ROI) encoding with PCRD optimization allows critical image areas (8×8 to 64×64 blocks) to be prioritized for lossless quality and early availability. Bitrate strategy is fully configurable, balancing image quality and compression ratio to meet mission needs.

Delivered as a configurable IP core for FPGAs, including Zynq 7000 & UltraScale+ family, with customizable generic parameters for flexible integration. The core operates at frequencies up to 200MHz with minimal resource utilization – 3,5% LUTs and 1% Flip-Flops on UltraScale+ ZU7EG. The implementation includes AXI4-Stream interfaces for seamless integration with existing satellite data processing pipeline and supports real-time compression with throughput rates up to 198.69 MSamples/s.

SKAIPACK is delivered as an optimized CPU binary or Python package, supporting both ARM and x86-64 architectures commonly found in satellite data handling units. Pre-configured ROI presets are included for common use cases like cloud screening and object detection. Memory requirements are typically between 50-200 MB, and processing rates can achieve 2-60 MSamples/s, depending on CPU performance and selected compression settings. Furthermore, the implementation incorporates advanced coregistration software with feature-based methods for accurate multispectral band alignment to enable even higher compression ratios.