SARA Thermal Reading

The SARA Thermal Reading provides an automated process to get the temerature in a chosen polygon in a thermal image.

Dependencies

The dependencies used for this package are listed in pyproject.toml and pinned in uv.lock. This ensures our builds are predictable and deterministic. This project uses uv for dependency management:

uv lock

To update the dependencies to the latest versions, run:

uv lock --upgrade

Setup

For the thermal reading to run, we need a reference image and reference polygon located in a blob store. Both the image and polygon needs to be stored in a container named installation_code and a folder named tagId_inspectionDescription. For an example see the saradev, saradevthermalref storage account.

Install locally

Install with uv sync --extra dev

Run tests

Run tests with uv run pytest .

Example .env.example

SOURCE_STORAGE_CONNECTION_STRING=DefaultEndpointsProtocol=ht ...
DESTINATION_STORAGE_CONNECTION_STRING=DefaultEndpointsProtocol=ht ...
REFERENCE_STORAGE_CONNECTION_STRING=ht ...

Dev utils

Create reference polygon

Draw polygon directly on image

python utils_cli.py create-polygon path/to/image.fff

Will by default save to reference_polygon.json in the current directory

Create reference polygon for thermal reference image

Draw polygon on thermal reference image, and save it to blob

python utils_cli.py create-polygon-cloud blobstorageaccountname "tagId" "inspectiondescription"

Show reference polygon for thermal reference image

Show reference polygon on thermal reference image

python utils_cli.py show-polygon-cloud blobstorageaccountname "tagId" "inspectiondescription"

Plot reference polygon local files

python utils_cli.py plot-fff path/to/image.fff --polygon-json-path path/to/reference_polygon.json

Plot cloud reference polygon

python utils_cli.py plot-current-reference-image-and-polygon \
    --installation-code "hua" --tag-id testtag --inspection-description testdesc

Run local fff workflow with example data

python utils_cli.py run-fff-workflow --polygon-path example-data/asset-example/polygon.json --reference-image-path example-data/asset-example/thermal_image.fff

Pipeline Diagram

graph TD
    subgraph Inputs
        A[Source Thermal Image]
        C[Reference Thermal Image]
        B[Reference Polygon .json]
    end

    subgraph Preprocessing
        D[Normalize temp range]
        E[Convert to uint8]
        F[CLAHE]
    end

    subgraph "Image Matching & Alignment"
        G["Phase Corr. + Hanning window"]
        H["Translation (dx, dy)"]
        J["Shift polygon by (dx, dy)"]
    end

    subgraph "Temperature Extraction"
        K["Create mask from polygon"]
        L[Apply mask to source image]
        M["Percentile(pixels, 95)"]
    end

    A --> D
    C --> D
    D --> E --> F
    F -->|"preprocessed source"| G
    F -->|"preprocessed reference"| G
    G --> H
    H --> J
    B --> J
    J --> K
    A -->|"raw float64 (°C)"| L
    K --> L --> M

    M --> N([P95 Temperature Output])

Loading