中国物理B ›› 2026, Vol. 35 ›› Issue (1): 14202-014202.doi: 10.1088/1674-1056/ade4b0

• • 上一篇    下一篇

Image-free single-pixel semantic segmentation for complex scene based on multi-scale U-Net

Tengfei Liu(刘腾飞), Yanfeng Bai(白艳锋)†, Jianxia Chen(陈健霞), Jintao Zhai(翟锦涛), Siqing Xiang(向思卿), Xianwei Huang(黄贤伟), and Xiquan Fu(傅喜泉)‡   

  1. College of Computer Science and Electronic Engineering, Hunan University, Changsha 410082, China
  • 收稿日期:2025-03-29 修回日期:2025-05-12 接受日期:2025-06-16 发布日期:2026-01-09
  • 通讯作者: Yanfeng Bai, Xiquan Fu E-mail:yfbai@hnu.edu.cn;fuxq@hnu.edu.cn
  • 基金资助:
    Project supported by the Fundamental Research Funds for the Central Universities of China (Grant No. 531118010757).

Image-free single-pixel semantic segmentation for complex scene based on multi-scale U-Net

Tengfei Liu(刘腾飞), Yanfeng Bai(白艳锋)†, Jianxia Chen(陈健霞), Jintao Zhai(翟锦涛), Siqing Xiang(向思卿), Xianwei Huang(黄贤伟), and Xiquan Fu(傅喜泉)‡   

  1. College of Computer Science and Electronic Engineering, Hunan University, Changsha 410082, China
  • Received:2025-03-29 Revised:2025-05-12 Accepted:2025-06-16 Published:2026-01-09
  • Contact: Yanfeng Bai, Xiquan Fu E-mail:yfbai@hnu.edu.cn;fuxq@hnu.edu.cn
  • Supported by:
    Project supported by the Fundamental Research Funds for the Central Universities of China (Grant No. 531118010757).

PDF (PC)

0

赞

摘要： Single-pixel imaging (SPI) receives widespread attention due to its superior anti-interference capabilities, and image segmentation technology can effectively facilitate its recognition and information extraction. However, the complexity of the target scene and plenty of imaging time in SPI make it challenging to achieve high-quality and concise segmentation. In this paper, we investigate the image-free intricate scene semantic segmentation in SPI. Using "learned" illumination patterns allows for the full extraction of the object's spatial information, thereby enabling pixel-level segmentation results through the decoding of the received measurements. Simulation and experimentation show that, in the absence of image reconstruction, the mean intersection over union (MIoU) of segmented image can reach higher than 85%, and the Dice coefficient (DICE) close to 90% even at the sampling ratio of 5%. Our approach may be favorable to applications in medical image segmentation and autonomous driving field.

关键词: Image-free single-pixel semantic segmentation for complex scene based on multi-scale U-Net

Abstract: Single-pixel imaging (SPI) receives widespread attention due to its superior anti-interference capabilities, and image segmentation technology can effectively facilitate its recognition and information extraction. However, the complexity of the target scene and plenty of imaging time in SPI make it challenging to achieve high-quality and concise segmentation. In this paper, we investigate the image-free intricate scene semantic segmentation in SPI. Using "learned" illumination patterns allows for the full extraction of the object's spatial information, thereby enabling pixel-level segmentation results through the decoding of the received measurements. Simulation and experimentation show that, in the absence of image reconstruction, the mean intersection over union (MIoU) of segmented image can reach higher than 85%, and the Dice coefficient (DICE) close to 90% even at the sampling ratio of 5%. Our approach may be favorable to applications in medical image segmentation and autonomous driving field.

Key words: single-pixel imaging, image-free, deep learning, complex scene

中图分类号:  (Imaging and optical processing)

  • 42.30.-d
42.30.Va (Image forming and processing) 42.30.Tz (Computer vision; robotic vision)