Integrated Optical Metrology System
DEPLOYED1. The Situation: Beyond the Human Eye
The Context: Validating mechanochromic materials requires more than just seeing a color change; it requires quantifying it. To prove a material is viable for injury monitoring, we need a precise correlation between mechanical strain and optical response in the CIE Lab* color space.
The Gap: Manual analysis was insufficient. High-speed video data (2000+ frames/test) suffered from lighting variances and mechanical “drift” (clamp slippage), making it impossible to separate true color shifts from environmental noise.
2. The Task: A “Ground Truth” Pipeline
I needed to architect a Computer Vision System capable of isolating the material’s physical response from experimental artifacts.
Key Requirements:
- Input Compliance: Automatically reject or correct video inputs (HDR/SDR) to ensure lighting standardization.
- Structural Integrity: Track the material’s texture (yarns) independent of the testing rig to eliminate mechanical noise.
- Rigorous Colorimetry: Quantify response using CIE Lab scalars, moving beyond basic RGB values to measure Perceptual Color Difference Delta E00/E76.
3. The Action: The approach
I developed a modular Python/OpenCV architecture that operates on two distinct analytical layers: Structural (Texture) and Scalar (Color Physics).
A. Gateway: Input Compliance & Pre-processing
Before analysis begins, the pipeline acts as a quality gate. It detects High-Dynamic-Range (HDR) profiles and uses an FFmpeg-based Tone Mapping engine (Hable method) to enforce a standardized SDR color profile. This ensures that every pixel analyzed represents true material reflectance, not camera auto-exposure artifacts.
B. Layer 1: Structural Analysis (Texture Tracking)
To solve the “drift” problem, I implemented ECC (Enhanced Correlation Coefficient) tracking. Instead of tracking the machine’s movement, the code locks onto the texture of the rigid clamps with sub-pixel accuracy. This creates a “Virtual Extensometer” that calculates strain based on the actual material deformation, removing fabric slippage noise entirely.
C. Layer 2: Scalar Analysis (CIE Colorimetry)
Once the region of interest is structurally stabilized, the pipeline performs deep colorimetry. It converts raw pixel data into CIE Lab coordinates—separating Lightness L* from Chromaticity a*, b*. This allows us to calculate Delta E00 (CIEDE2000), providing a mathematically rigorous metric for “Visible Color Change” that matches human perception.
(HDR/MOV)"/]:::input subgraph "The Compliance Gateway" Check{"HDR Check"}:::gateway Norm["Standardization
(Hable Tone Mapping)"]:::gateway end subgraph "Layer 1: Structural Analysis" ECC["Sub-Pixel ECC Tracking
(Texture Locking)"]:::structure Drift["Zero-Drift Compensation"]:::structure end subgraph "Layer 2: Scalar Analysis (CIE)" Lab["RGB to CIE L*a*b*
Conversion"]:::scalar Metrics["Calculate ΔE2000
(Perceptual Difference)"]:::scalar PCA["Feature Ranking
(Signal Isolation)"]:::scalar end Out[/"Quantified Color
Response Curves"/]:::input %% Flow Video --> Check Check -- "Compliance Fail" --> Norm Norm --> ECC Check -- "Pass" --> ECC ECC --> Drift Drift --> Lab Lab --> Metrics Metrics --> PCA PCA --> Out
4. The Result
- Scientific Precision: Successfully decoupled mechanical noise from optical signal, revealing a “Hysteresis Loop” in the color response (fatigue history) that was previously invisible in manual analysis.
- Throughput: Reduced data processing time by >90% (from 4 hours/video to 5 minutes), enabling high-volume statistical validation.
- Standardization: This pipeline is now the primary validation tool, providing the quantitative CIE metrics required to benchmark new mechanochromic materials against known literature work.
