🔗 🇧🇷 Read this page in Portuguese

LightSense-Calib: Spectral Calibration of the AS7341 Sensor

📌 Project Overview

This project is part of my Undergraduate Thesis (TCC) at UFSM, focusing on the calibration of a multi-channel light sensor, specifically the OSRAM AS7341. More details can be found in my thesis: UFSM Repository.

🔧 AS7341 Sensor

Below is an image of the AS7341 sensor used in this project:

---

🎯 Objective

The goal is to calibrate the AS7341 sensor by comparing its reconstructed spectral data with high-fidelity reference spectra obtained through an integrating sphere. The calibration algorithms aim to determine the best set of coefficients that minimize the error between the sensor's reconstructed spectrum and the reference spectrum.

---

🛠️ Hardware Used

The experimental setup consists of the following components:

• Sensor: OSRAM AS7341 (multi-channel spectral sensor)
• Microcontroller: Arduino (for reading and transmitting data via serial communication)
• Integrating Sphere: Used as a high-precision spectral reference
• Computer: Laptop running MATLAB for data processing

Image of the experimental setup, including the sensor inside the integrating sphere, the laptop running MATLAB, and other experiment components:

---

🔬 Calibration Process

The project includes three main MATLAB scripts:

1️⃣ channels_calib_no_diff.m

• This script calibrates each of the 9 spectral channels of the AS7341 sensor.

• Generates calibration coefficients for later use.

• Produces graphs and images detailing each step of the process.

  

2️⃣ channels_calib_with_diff.m

• Uses the coefficients from the previous calibration.

• Calibrates the sensor with a diffuser, essential for ensuring accurate light measurements.

• Generates the final coefficients considering the diffuser.

  

3️⃣ sensor_processor.m

• Receives sensor data and processes it with or without calibration.

• Displays measurements such as Irradiance, Illuminance, and PPFD (Photosynthetic Photon Flux Density).

  

---

🏆 Calibration Methodology

Calibration data is obtained by comparing:

1. The reconstructed spectrum from the AS7341 sensor.
2. The reference spectrum of the same light source, measured inside an integrating sphere (a high-precision instrument).

The algorithms optimize the calibration coefficients to minimize errors between these two measurements.

📊 AS7341 Sensor Calibration Workflow

The image below illustrates the sensor calibration process, highlighting the steps of coefficient optimization, R² validation, and result comparison with an integrating sphere:

---

📈 Evaluation Metrics

To assess calibration quality, the coefficient of determination (R²) was used to indicate how well the calibrated spectra fit the reference spectrum. Methods such as Mean Squared Error (MSE), Mean Absolute Error (MAE), and Weighted Mean Squared Error (WMSE) were analyzed to determine the best calibration coefficients.

---

📜 More Information

For a detailed explanation of the methodology, results, and implementation, refer to my thesis: UFSM Repository.