SLSong Li

Song Li

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Song Li has a total of 7 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Suspended Insulation Structure Design for Infrared Thermal Detector

    by Song Li, Yufei Zhai, Yuxuan Dong, Jianqing Cai, Min Wang

    Infrared thermal detectors generate signal output utilizing thermal effect, with detection performance being dictated by the structure design of the detector. To ensure effective thermal insulation, detectors are typically designed with slender supporting beams to reduce thermal conductivity and enhance signal output. In this paper, a fluorescent infrared detector with different isolation beam designs is proposed to investigate the trade-off relationship among detector performance parameters. Two kinds of isolation beams are theoretically derived to minimize the thermal conductivity for specific detection unit, and the thermal detectors with suspending units are manufactured by MEMS technology. The thermal imaging results for a 623 K heat source indicate that the temperature rise for the two-beam structure (i.e. 46.7 K) exceeds that of the four-beam structure (i.e. 38.3 K). Additionally, the detectivity of 4.05 $\times$ 10 $^{\mathbf{7}}$ cm $\cdot$ Hz $^{\mathbf{1/2}}$ /W is obtained by the two-beam structure, which is lower than that of the four-beam structure (i.e. 5.80 $\times$ 10 $^{\mathbf{7}}$ cm $\cdot$ Hz $^{\mathbf{1/2}}$ /W). The temporal resolution and NETD are also calculated and compared. The findings demonstrate that in designing thermal detectors, deliberately sacrificing a portion of the detectivity within an acceptable range and reducing beam thermal conductivity can significantly enhance temperature rise and increase signal output. 2024-0131


  • Privacy-Preserving Infrared Remote Sensing System for Human Presence Detection

    by Yuxuan Dong, Song Li, Yufei Zhai, Zhuoyu Chen, Fengwei An, Min Wang

    The infrared (IR) remote sensing system comprised of a fluorescence-based thermal detector and image processing algorithm is proposed in this paper. Instead of column readout architectures for integrated circuits for microbolometers, the fluorescent IR detector governed by infrared-to-visible conversion can generate the thermal image directly, which can be easily processed by the back-end visible image sensor. Such architecture allows separate modulation of thermal conversion and signal processing. To improve the detectivity of the system, image processing algorithms are proposed to achieve robust quasi-real-time human detection. Adjustable thresholds and edge enhance operators are combined to extract human candidate regions, which are compatible with commercial FPGA. The experimental results indicate that the noise intensity in the subtracted thermal image has been reduced by 98% compared to the non-local means method or bilateral method. A 5.8-fold improvement in terms of signal-to-noise ratio (SNR) is exhibited in long-wavelength human radiation. The detectivity of the proposed detector was improved from 5.06 × 10 6 cm·Hz 1/2 /W to 4.33 ×10 8 cm·Hz 1/2 /W after being equipped with optimized image conditioning circuit. A quasi-real-time image processing verification platform was successfully established in Xilinx FPGA, providing a 16-bit 1360 × 1024 resolution grayscale video stream output at 37 fps. Our work demonstrates the capability of the IR detection system in human activity detection, indicating its potential in human action recognition applications.