University Honors

Insect monitoring is important for agriculture, biodiversity research, and public health, but many traditional monitoring methods can be expensive, labor-intensive, destructive, or difficult to apply over large areas. Acoustic insect monitoring offers a promising alternative because insects can be detected by analyzing the unique sounds produced by their wingbeats. However, real-world audio recordings collected by sensors often include background noise, environmental interference, and other non-insect sounds, making it difficult to accurately identify true insect activity. The goal of this project was to optimize acoustic insect sensor performance by developing a MATLAB-based workflow that can process insect audio recordings and identify sound segments most likely to contain insect wingbeat signals. The workflow incorporated frequency-domain signal processing techniques including Fast Fourier Transform (FFT) analysis, dominant-frequency extraction, spectral filtering, area-under-the-curve (AUC) scoring, and sliding-window subsequence detection to isolate insect-like acoustic patterns from noisy recordings. Rather than relying only on raw recordings, this project focused on improving the preprocessing and signal-analysis steps that occur before insect classification. By making these steps more reliable, the sensor can better separate useful insect data from unwanted noise and produce more accurate wingbeat frequency information. This work provides a foundation for developing acoustic insect monitoring systems that are more accurate, scalable, and practical for real-world use. With further testing and development, this type of sensor-based system could support improved pest management, ecological monitoring, and public health efforts related to disease-carrying insects.