
浏览全部资源
扫码关注微信
1.南京信息工程大学 自动化学院, 江苏 南京 210044
2.江苏省大气环境与装备技术协同创新中心, 江苏 南京 210044
严飞,男,博士,副教授,2013年于中国科学院长春光学精密机械与物理研究所获得博士学位,主要从事三维测量与显示、嵌入式图像处理技术等方面的研究。E-mail:fyan@nuist.edu.cn
刘银萍,女,博士,副教授,2013年于吉林大学获得博士学位,主要从事电磁场计算、安全技术等方面的研究。E-mail:yinpingliu@nuist.edu.cn
收稿:2026-07-23,
修回:2026-08-20,
纸质出版:2026-09-05
移动端阅览
严飞, 邵俊楠, 王翔宇, 等. 基于ZYNQ与YOLO26n的PCB缺陷检测[J]. 液晶与显示, 2026,41(9):1315-1324.
YAN Fei, SHAO Junnan, WANG Xiangyu, et al. PCB defect detection based on ZYNQ and YOLO26n[J]. Chinese Journal of Liquid Crystals and Displays, 2026, 41(9): 1315-1324.
严飞, 邵俊楠, 王翔宇, 等. 基于ZYNQ与YOLO26n的PCB缺陷检测[J]. 液晶与显示, 2026,41(9):1315-1324. DOI: 10.3724/CJLCD.2026-0138. CSTR: 32172.14.CJLCD.2026-0138.
YAN Fei, SHAO Junnan, WANG Xiangyu, et al. PCB defect detection based on ZYNQ and YOLO26n[J]. Chinese Journal of Liquid Crystals and Displays, 2026, 41(9): 1315-1324. DOI: 10.3724/CJLCD.2026-0138. CSTR: 32172.14.CJLCD.2026-0138.
印制电路板(Printed Circuit Board,PCB)作为电子设备中的基础部件,其表面缺陷会影响电子产品的可靠性与使用寿命。针对PCB缺陷检测模型端侧部署中计算量大、存储开销高及功耗受限等问题,本文设计了一种基于ZYNQ SoC平台与YOLO26n目标检测模型的PCB缺陷检测系统。针对卷积计算瓶颈,本文采用Vitis HLS设计卷积加速IP核,并结合空间分块、通道并行、循环展开及双缓冲乒乓调度等策略提高计算并行度与数据传输效率;同时采用16位Q7.9定点量化及扩展位宽累加降低数据存储与传输开销。实验结果表明,全网络170个算子均完成浮点与定点路径对齐,卷积拓扑和Q7.9定点路径的余弦相似度最小值分别为0.989 4和0.999 8。CPU浮点与ZYNQ定点推理在各类缺陷检测指标上整体保持接近,精度(Precision)、召回率(Recall)和mAP@0.5最大差值分别为0.023、0.024和0.012。硬件资源中查找表(Look-Up-Table,LUT)、触发器(Flip-Flop,FF)及数字信号处理器(DSP)使用率分别为15.42%、7.30%和20.64%,整体功耗为2.615 W,表明系统能够在保持较高检测一致性的同时实现低功耗端侧推理。
Printed Circuit Boards(PCBs) are fundamental components in electronic devices, and surface defects can affect the reliability and lifespan of electronic products. To address the issues of high computational cost, high storage overhead, and limited power consumption in edge deployment of PCB defect detection models, this paper designs a PCB defect detection system based on the ZYNQ SoC platform and the YOLO26n target detection model. To overcome the bottleneck of convolutional computation, this paper employs a Vitis HLS convolution acceleration IP core and combines spatial partitioning, channel parallelism, loop unrolling, and double-buffered ping-pong scheduling strategies to improve computational parallelism and data transmission efficiency. Simultaneously, 16-bit Q7.9 fixed-point quantization and extended bit-width accumulation are used to reduce data storage and transmission overhead. Experimental results show that all 170 operators in the network achieve floating-point and fixed-point path alignment, and the minimum cosine similarity between the convolutional topology and the Q7.9 fixed-point path are 0.989 4 and 0.999 8, respectively. CPU floating-point and ZYNQ fixed-point inference maintained similar performance across various defect detection metrics, with the largest differences in Precision, Recall, and mAP@0.5 being 0.023, 0.024, and 0.012, respectively. Hardware resource utilization rates of Look-Up-Table (LUT), Flip-Flop (FF), and Digital signal Processor (DSP) are 15.42%, 7.30%, and 20.64%, respectively, with an overall power consumption of 2.615 W. This indicates that the system can achieve low-power edge inference while maintaining high detection consistency.
吴一全 , 赵朗月 , 苑玉彬 , 等 . 基于机器视觉的PCB缺陷检测算法研究现状及展望 [J]. 仪器仪表学报 , 2022 , 43 ( 8 ): 1 - 17 .
WU Y Q , ZHAO L Y , YUAN Y B , et al . Research status and the prospect of PCB defect detection algorithm based on machine vision [J]. Chinese Journal of Scientific Instrument , 2022 , 43 ( 8 ): 1 - 17 . (in Chinese)
徐一奇 , 肖金球 , 汪俞成 , 等 . 基于机器视觉的PCB表面缺陷检测研究综述 [J]. 微电子学与计算机 , 2025 , 42 ( 4 ): 1 - 15 .
XU Y Q , XIAO J Q , WANG Y C , et al . A review of research on PCB surface defect detection based on machine vision [J]. Microelectronics & Computer , 2025 , 42 ( 4 ): 1 - 15 . (in Chinese)
PARK J , CHOI Y , KWON S , et al . Microsphere-assisted hyperspectral imaging: super-resolution, non-destructive metrology for semiconductor devices [J]. Light: Science & Applications , 2024 , 13 ( 1 ): 122 . doi: 10.1038/s41377-024-01469-3 http://dx.doi.org/10.1038/s41377-024-01469-3
刘智 , 赖庆荣 , 张天禹 , 等 . 激光多普勒测振技术无损检测果冻橙粒化病 [J]. 中国光学(中英文) , 2025 , 18 ( 1 ): 160 - 169 . doi: 10.37188/co.2024-0115 http://dx.doi.org/10.37188/co.2024-0115
LIU Z , LAI Q R , ZHANG T Y , et al . Non-destruction detection of jelly orange granulation disease using laser Doppler vibrometry [J]. Chinese Optics , 2025 , 18 ( 1 ): 160 - 169 . (in Chinese) . doi: 10.37188/co.2024-0115 http://dx.doi.org/10.37188/co.2024-0115
ZHANG Z L , KONG L Y , ZHANG L H Y , et al . Structured light meets machine intelligence [J]. eLight , 2025 , 5 ( 1 ): 26 . doi: 10.1186/s43593-025-00102-z http://dx.doi.org/10.1186/s43593-025-00102-z
GIRSHICK R , DONAHUE J , DARRELL T , et al . Rich feature hierarchies for accurate object detection and semantic segmentation [C]. 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus : IEEE , 2014 : 580 - 587 . doi: 10.1109/cvpr.2014.81 http://dx.doi.org/10.1109/cvpr.2014.81
REN S Q , HE K M , 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 . doi: 10.1109/tpami.2016.2577031 http://dx.doi.org/10.1109/tpami.2016.2577031
REDMON J , DIVVALA S , GIRSHICK R , et al . You only look once: unified, real-time object detection [C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition . Las Vegas : IEEE , 2016 : 779 - 788 . doi: 10.1109/cvpr.2016.91 http://dx.doi.org/10.1109/cvpr.2016.91
BOCHKOVSKIY A , WANG C Y , LIAO H Y M . YOLOv4: optimal speed and accuracy of object detection [J/OL]. arXiv , 2020 : 2004 . 10934 . doi: 10.48550/arXiv.2004.10934 http://dx.doi.org/10.48550/arXiv.2004.10934
TANG J L , LIU S B , ZHAO D X , et al . PCB-YOLO: an improved detection algorithm of PCB surface defects based on YOLOv5 [J]. Sustainability , 2023 , 15 ( 7 ): 5963 . doi: 10.3390/su15075963 http://dx.doi.org/10.3390/su15075963
XIAO G S , HOU S L , ZHOU H Y . PCB defect detection algorithm based on CDI-YOLO [J]. Scientific Reports , 2024 , 14 ( 1 ): 7351 . doi: 10.1038/s41598-024-57491-3 http://dx.doi.org/10.1038/s41598-024-57491-3
YUAN M H , ZHOU Y B , REN X Y , et al . YOLO-HMC: an improved method for PCB surface defect detection [J]. IEEE Transactions on Instrumentation and Measurement , 2024 , 73 : 2001611 . doi: 10.1109/tim.2024.3351241 http://dx.doi.org/10.1109/tim.2024.3351241
朱黎颖 , 王森 , 沈爱萍 , 等 . 复杂背景下的电路板表面焊接缺陷视觉检测 [J]. 光学 精密工程 , 2024 , 32 ( 14 ): 2256 - 2271 . doi: 10.37188/ope.20243214.2256 http://dx.doi.org/10.37188/ope.20243214.2256
ZHU L Y , WANG S , SHEN A P , et al . Visual inspection of soldering defects on board surfaces against complex backgrounds [J]. Optics and Precision Engineering , 2024 , 32 ( 14 ): 2256 - 2271 . (in Chinese) . doi: 10.37188/ope.20243214.2256 http://dx.doi.org/10.37188/ope.20243214.2256
杨彦萍 , 高军伟 , 刘兆龙 , 等 . 基于改进YOLO11的印刷电路板表面缺陷检测算法 [J]. 液晶与显示 , 2025 , 40 ( 10 ): 1545 - 1556 . doi: 10.37188/cjlcd.2025-0137 http://dx.doi.org/10.37188/cjlcd.2025-0137
YANG Y P , GAO J W , LIU Z L , et al . Surface defect detection on printed circuit boards based on improved YOLO11 [J]. Chinese Journal of Liquid Crystals and Displays , 2025 , 40 ( 10 ): 1545 - 1556 . (in Chinese) . doi: 10.37188/cjlcd.2025-0137 http://dx.doi.org/10.37188/cjlcd.2025-0137
高雪 , 袁飞 , 霍艳飞 . 基于改进DEIM的印刷电路板缺陷检测算法 [J]. 液晶与显示 , 2026 , 41 ( 5 ): 700 - 716 .
GAO X , YUAN F , HUO Y F . PCB defect detection algorithm based on improved DEIM [J]. Chinese Journal of Liquid Crystals and Displays , 2026 , 41 ( 5 ): 700 - 716 . (in Chinese)
DONG Y B , BAI Y C , ZHANG Q M , et al . High-throughput optical neuromorphic graphic processing at millions of images per second [J]. eLight , 2025 , 5 ( 1 ): 29 . doi: 10.1186/s43593-025-00106-9 http://dx.doi.org/10.1186/s43593-025-00106-9
MELONI P , CAPOTONDI A , DERIU G , et al . NEURAghe: exploiting CPU-FPGA synergies for efficient and flexible CNN inference acceleration on Zynq SoCs [J]. ACM Transactions on Reconfigurable Technology and Systems , 2018 , 11 ( 3 ): 18 . doi: 10.1145/3284357 http://dx.doi.org/10.1145/3284357
JOCHER G , QIU J , LIU M Y , et al . Ultralytics YOLO26: unified real-time end-to-end vision models [J/OL]. arXiv , 2026 : 2606 . 03748 .
TANG S L , HE F , HUANG X L , et al . Online PCB defect detector on a new PCB defect dataset [J/OL]. arXiv , 2019 : 1902 . 06197 .
DAI D Y , ZHANG Y C , ZHANG J H , et al . Trainable fixed-point quantization for deep learning acceleration on FPGAs [J/OL]. arXiv , 2024 : 2401 . 17544 .
0
浏览量
51
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621