signal-digitizer

signal-digitizer is a Python library for converting scanned charts and grid-based plots into calibrated one-dimensional signals. It is designed for applications such as ECG traces, laboratory recorder outputs, and other plotted signals stored as PDF documents.

The library processes a chart by rendering the page, correcting skew, detecting the grid, extracting the signal trace, and converting pixel coordinates into calibrated signal values.

Installation

pip install signal-digitizer

Quick Start

import signal_digitizer as sd

x, y = sd.digitize(
    "chart.pdf",
    unit_per_vgap=1.0,
    unit_per_hgap=1.0,
)

The returned x and y arrays contain the extracted and calibrated signal coordinates.

Command Line Usage

signal-digitizer chart.pdf -o signal.csv

This processes the input PDF and saves the extracted signal to a CSV file.

How It Works

The digitization workflow consists of the following stages:

1. Page rendering – The input PDF page is converted into an image.
2. Skew correction – Page rotation is estimated and corrected.
3. Grid detection – Grid lines are identified using image-processing techniques.
4. Calibration – Grid spacing is used to convert pixel coordinates into data units.
5. Trace extraction – The plotted signal is separated from the chart image.
6. Signal generation – The extracted trace is converted into calibrated x and y values.

Main API

digitize()

Processes a PDF file through the complete signal digitization pipeline.

x, y = sd.digitize(
    "chart.pdf",
    unit_per_vgap=1.0,
    unit_per_hgap=1.0,
)

digitize_page()

Processes an already opened PyMuPDF page.

import pymupdf
import signal_digitizer as sd

document = pymupdf.open("chart.pdf")

x, y = sd.digitize_page(
    document[0],
    unit_per_vgap=1.0,
    unit_per_hgap=1.0,
)

digitize_image()

Processes an image that has already been loaded as a NumPy array.

x, y = sd.digitize_image(
    image_array,
    unit_per_vgap=1.0,
    unit_per_hgap=1.0,
)

Core Functionality

The library includes functionality for:

• PDF page rendering
• Skew estimation and correction
• Grid-line detection
• Axis calibration
• Signal trace extraction
• Conversion of extracted traces into calibrated signals
• Signal validation using Pearson correlation and RMSE

Signal Validation

If a reference signal is available, the extracted signal can be evaluated using:

result = sd.validate_signal(y, reference_signal)
print(result)

The validation result includes correlation and error-based measures for comparing the extracted signal with the reference.

Requirements

signal-digitizer requires Python 3.9 or later.

Its core dependencies include:

• PyMuPDF
• NumPy
• OpenCV
• SciPy

These dependencies are installed automatically with the package.

Authors

• Manoj Kumar C S
• V N Manjunath Aradhya
• Nikhil D Bharadwaj

Maintainers

• Manoj Kumar C S
• Nikhil D Bharadwaj

License

This project is distributed under the MIT License. See the LICENSE file for the complete license text.