SIH 2026 • PS ID: 26166 • SPACE TECHNOLOGY

Spatial Photometric Alignment
& Registration Kernel

An innovative automated software tool by Team byteforce designed to align Chandrayaan-2 optical images (OHRC, TMC, IIRS) with lunar reference maps with sub-pixel accuracy. Completely sun-angle and scale invariant.

Watch Product Demo View Repository

Mission Control: Interactive Prototype

The SPARK Pipeline

1. Illumination Normalization

Handles extreme sun-angle changes using Multi-Scale Retinex (15, 80, 250) + CLAHE for a sun-angle invariant representation.

2. Multi-Scale Search

Combines Dynamic Scale Pyramids (4x to 1/8x) with Affine-SIFT to achieve complete viewpoint and scale invariance.

3. Robust Matching

Utilizes MAGSAC homography estimation coupled with Spatial Grid Bucketing to guarantee uniformly distributed inliers.

4. Sub-Pixel Refinement

Implements CornerSubPix homography to push matching precision beyond the pixel level for ultimate accuracy.

5. Geometric Transform

Applies Homography (USAC MAGSAC) with an intelligent fallback to TPS (Thin Plate Spline) for optional non-rigid alignment.

6. Validation & Outputs

Generates detailed metrics (RMSE, SSIM, MI) alongside interactive 3D Topographic visualization and mission-ready PDF reports.

Powered By Modern Tech

Mature, open-source, and space-proven frameworks.

Python 3.10+
OpenCV
NumPy
Scikit-Learn
Plotly
Streamlit

Impact & Viability

Sub-Pixel

High Accuracy

Low RMSE

Robust & Reliable

Future Ready

Modular Architecture

Cost-Efficient

Open-Source Stack

Target Audience: ISRO & Space Agencies, Researchers, and Academia.
SPARK empowers India's leadership in lunar science by automating cross-sensor image alignment for precise mapping, analysis, and mission planning.