计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (6A): 240600069-6.doi: 10.11896/jsjkx.240600069

• 图像处理&多媒体技术 • 上一篇    下一篇

激光透窗低质量成像人体目标检测算法

伍智华, 程江华, 刘通, 蔡亚辉, 程榜, 潘乐昊   

  1. 国防科技大学电子科学学院 长沙 410073
  • 出版日期:2025-06-16 发布日期:2025-06-12
  • 通讯作者: 伍智华(18627599098@163.com)
  • 基金资助:
    湖南省自然科学基金(2020JJ4670)

Human Target Detection Algorithm for Low-quality Laser Through-window Imaging

WU Zhihua, CHENG Jianghua, LIU Tong, CAI Yahui, CHENG Bang, PAN Lehao   

  1. College of Electronic Science and Technology,National University of Defense Technology,Changsha 410073,China
  • Online:2025-06-16 Published:2025-06-12
  • About author:WU Zhihua,born in 1992,master.His main research interests include intelligent image processing and so on.
  • Supported by:
    Natural Science Foundation of Hunan Province,China(2020JJ4670).

PDF (PC)

35

摘要： 针对激光透窗低质量成像下的人体目标检测,现有算法存在检测不准确、识别率低等问题,提出一种基于YOLOv8n优化改进的目标检测算法YOLO-TC。重新设计主干部分的特征提取模块,提升模型多尺度特征表示能力;对YOLOv8n模型做剪枝处理,优化网络结构,降低模型复杂度的同时提升检测精度;在C2f模块与解耦头(Detect)之间添加EMA注意力机制模块,增强特征融合中的语义和位置信息,提升模型的特征融合能力;使用SIoU边界框回归损失函数代替原损失函数,提升算法推理的准确性和训练速度。实验结果表明,改进后的模型在激光透窗成像数据集中的精确度(Precision)、召回率(Recall)和平均精度均值(mAP)相比原模型分别提高了7.7%,5.9%和7.0%,模型大小缩减了34.6%,便于后续边缘端的硬件部署。

关键词: 激光透窗成像, YOLOv8, 多尺度特征提取, 模型剪枝, 注意力机制, SIoU

Abstract: In response to the challenges of inaccurate detection and low recognition rates in human target detection under low-quality imaging with laser through-window technology,an enhanced target detection algorithm,YOLO-TC,based on YOLOv8n optimization,has been proposed.The feature extraction module of the backbone has been redesigned to enhance the model’s multi-scale feature representation capability.Pruning of the YOLOv8n model has been employed to optimize the network structure,reduce model complexity,and enhance detection accuracy.An EMA attention mechanism module has been introduced between the C2f module and the detection head(Detect) to improve semantic and location information in feature fusion and enhance the model’s feature fusion ability.Using SIoU bounding box regression loss function instead of the original loss function to improve the inference accuracy and training speed of the algorithm.Experimental results on a laser through-window imaging dataset demonstrate that the Precision,Recall,and mean Average Precision(mAP) of the improved model has increased by 7.7%,5.9%,and 7.0% respectively.Furthermore,the model size has been reduced by 34.6% compared to the original model,making it suitable for subsequent edge hardware deployment.

Key words: Laser through-window imaging, YOLOv8, Multi-scale feature extraction, Model pruning, Attention mechanism, SIoU

中图分类号: 

  • TP391
[1]HU Y H,ZHAO L D.Research status and prospect of laser imaging processing technology[J].InfraredandLaser Engineering,2023,52(6):9-29.
[2]ZHOU L X,HAN X D,YE S W,et al.Efficiency testing method for the echo receiving system of laser ranging station[J].Optics and Lasers in Engineering,2024,176:108061.
[3]YANG A F.Application of Short-wave infrared laser in photoelectric reconnaissance and anti-reconnaissance[J].Applied Optics,2019,40(6):937-943.
[4]JIANG T.Research on high-speed through-window imagingtechnology of dynamic target under strong reflection interfe-rence[D].Beijing:University of Science and Technology Beijing,2021.
[5]ZHANG S.Research on target detection algorithm based onconvolutional neural network under laser active imaging [D].University of Chinese Academy of Sciences,2021.
[6]LI Z T,CHU Y X,CHEN Y.Fire and smoke detection technology based on laser range gating imaging[J].Fire Science and Technology,2023,42(9):1201-1204.
[7]WANG S,PAN Y Z,LIU Y,et al.Research on improving image quality of laser active imaging in fog[J].Infrared and Laser Engineering,2013,42(9):2392-2396.
[8]ZHAO Z J,TAN Y G,LIU P,et al.Research on Intelligent recognition system of underwater long-range target based on laser gating imaging technology[J].Integration Technology,2023,12(2):39-52.
[9]REDMON J,DIVVALA S,GIRSHICK R,et al.You onlylookonce:unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016:779-788.
[10]DAVID A,DANIEL M,DROR R,et al.Pressure estimation via measurement of reduced light scattering coefficient by oblique laser incident reflectometry[J].Journal of Laser Applications,2024,36(1):012028.
[11]SHANGGAUN M,LIAO Z Y,GUO Y R,et al.Sensing the profile of particulate beam attenuation coefficient through a single-photon oceanic Raman lidar[J].Optics Express,2023,31(16):25398-25414.
[12]PANG Z H,SONG C T,LIU B H.A study on accurate ranging method of a dual-wavelength orthogonal FMCW laser fuze in a complex aerosol environment[J].IEEE Sensors Journal,2024,24(7):113.6-113.15.
[13]ANDREA T,BORIS M.Robust attenuated total reflection infrared spectroscopic sensors based on quantum cascade lasers for harsh environments[J].IEEE Sensors Journal,2024,24(1):814-821.
[14]ROSS G,JEFF D,TREVOR D,et al.Region-Based Convolutional Networks for Accurate Object Detection and Segmentation[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,2016,38(1):142-158.
[15]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:Towards Real-Time Object Detection with Region Proposal Networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
[16]KAIMING H,GEORGI G,PIOTR D,et al.Mask R-CNN[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,2020,42(2):386-397.
[17]LIU W,ANGUELOV D,ERHAN D,et al.SSD:single shotmultibox detector[C]//Computer Vision-ECCV 14th European Conference.Amsterdam,The Netherlands,2016:21-37.
[18]TERVEN J,CORDOVA-ESPARZA D.A Comprehensive Re-view of YOLO:From YOLOv1 and Beyond[J].arXiv:2304.00501,2023.
[19]LIU Z,LI J G,SHEN Z Q,HUANG G,et al.Learning Efficient Convolutional Networks through Network Slimming[C]//IEEE International Conference on Computer Vision.Venice,Italy,2017:2755-2763.
[20]OUYANG D L,HE S,ZHANG G Z,et al.Efficient multi-scale attention module with cross-spatial learning[C]//IEEE International Conference on Acoustics,Speech and Signal Processing.Rhodes Island,Greece,2023:1-5.
[21]GEVORGYAN Z.SIoU loss:more powerful learning for bounding box regression[J].arXiv:2205.12740,2022.
[22]XIE Y M,ZHANG L W,YU X Y,et al.YOLO-MS:Multispectral Object Detection via Feature Interaction and Self-Attention Guided Fusion[J].IEEE Transactions on Cognitive and Deve-lopmental Systems,2023,15(4):2132-2143.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发