Image-Based Volatile Organic Compound Identification Using the Cosine Similarity Method.

Area of Science:

• Analytical Chemistry
• Environmental Science
• Sensor Technology

Background:

• Accurate detection and identification of volatile organic compounds (VOCs) are crucial for environmental monitoring, industrial safety, and human health.
• Current microsensor techniques for VOC detection primarily use electrical or frequency signals, limiting chemical identification capabilities.

Purpose of the Study:

• To develop a visual and intuitive method for identifying VOCs.
• To demonstrate the potential of plasma image analysis for VOC identification.

Main Methods:

• Utilized a helium-based dielectric plasma detector to photoionize gas samples.
• Recorded plasma color changes upon VOC injection using a standard phone camera.
• Employed cosine similarity analysis on segmented plasma images for VOC identification.

Main Results:

• Achieved nearly 100% accuracy in identifying VOCs (n-pentane, benzene, acetone) and their mixtures.
• Demonstrated the effectiveness of cosine similarity on direct image segmentation for VOC identification.
• Identified sensitivity to specific image regions impacting identification accuracy.

Conclusions:

• Plasma image analysis offers a new, accurate approach for VOC identification.
• The developed method shows potential for creating low-cost, visually intuitive VOC sensors.
• This technique overcomes limitations of traditional electrical or frequency-based VOC detection methods.