Fiber-optic ultrasound sensors hold considerable promise for diverse applications, benefiting from their high sensitivity and advantageous physical characteristics. However, their practical adoption is often hindered by the operational instability of conventional high-speed demodulation techniques, which rely on sideband filtering and require active quadrature-point locking. To address this fundamental constraint, this work introduced a high-sensitivity Fabry-Pérot (FP) fiber-optic ultrasound sensing system incorporating an optoelectronic oscillator. The system integrated a submicron-precision Fabry-Pérot microcavity, fabricated on the fiber tip via two-photon polymerization (TPP), with a low-phase-noise optoelectronic oscillator (OEO) of excellent frequency stability. Experimental characterization demonstrated ultrasonic sensitivity of 3.089 mV/kPa at 1 MHz and low noise-equivalent pressure of 168 Pa, achieved without high-reflectivity coatings. By merging fiber-optic sensing with microwave photonics, the proposed architecture eliminated the need for active sideband locking, thereby substantially improving operational stability in complex and dynamic environments. This advancement establishes a reliable technical pathway for precision applications including structural defect mapping, intravascular photoacoustic imaging, and underwater acoustic holography.

Emotion recognition is vital for mental and physical health, but current methods often rely on active user participation (e.g., questionnaires) or wearable sensors, limiting their real-world adoption. This paper proposes a non-contact ballistocardiogram (BCG) based cross-subject emotion recognition framework. Subject-specific data distribution variations challenge model generalizability across domains. For this reason, we propose an LSCS-BDA method, extending Balanced Distribution Adaptation (BDA) with a Label Similarity Clustering for Selection (LSCS) module to optimize source domain selection. After feature extraction and baseline normalization, only the source domain individuals that are more similar to the target are included in the final source domain set through our method’s LSCS module (reducing training data to 9.78%), and finally the transfer is completed by the BDA algorithm. The experimental results show that the classification accuracy of our proposed method for three emotions of positive, negative, and neutral is 70.38%, which achieves higher accuracy and faster model training than classical transfer algorithms. The LSCS module in our method can be extended to other transfer learning methods. Our approach offers a new idea for contactless emotion recognition based on BCG and contributes to advancing its applications.