Geometric and Radiometric Image Data Resampling

GRIDR is a library for resampling and filtering raster image data, designed for efficiency in both in-memory processing and I/O operations.

Functional Scope & Features

Core Capabilities

• Grid-based Resampling
  • Adapt raster data to a target geometry defined by a grid containing the coordinates of each target pixel in the source image geometry.
  • Supports both full-resolution and under-sampled resolution grids.
  • Interpolation Methods : Nearest neighbor, linear, cubic
  • Mask Support:
    • Grid Masks: Raster or sentinel values.
    • Source Image Masks: Raster, sentinel values, or vectorized geometry.
    • Target Mask Production: Generate masks for the target raster geometry.
• Filtering: Apply spatial filters in the frequency domain (e.g., low-pass filtering).
• Mask Rasterization: Convert vectorized geometry masks into a regular target raster geometry.
• Optimized Workflows: Reduce I/O overhead for large-scale processing.

Function Types

1. Elemental (Core) Functions
   • Standalone operations for direct manipulation of in-memory data.
   • Ideal for custom processing pipelines and fine-grained control.
2. Chained Functions
   • Optimized sequences of operations to minimize I/O overhead.
   • Efficiently manage memory and CPU usage for large-scale processing.

Technical Implementation

Architecture

• Python: Core functionality and interface (not just for bindings).
• Rust: Performance-critical algorithms and heavy computations.
• PyO3: Used for seamless Python-Rust bindings.

Key Technical Aspects

• Rust Core Library: Can be used independently in other Rust projects.
• Python Integration: Full-featured methods available in Python, not just bindings.
• Optimized I/O: Designed to handle large datasets efficiently.

Getting Started

To install and use GRIDR, refer to the online documentation