A projectile targeting coordinate detection algorithm was proposed based on the Transformer architecture to address the problem of insufficient positioning accuracy in complex occlusion scenarios in photoelectric coordinate detection systems. The algorithm combined FPGA for signal acquisition and preprocessing, and leveraged the powerful feature extraction capability of the Transformer to optimize photoelectric coordinate detection accuracy. The system used a 1 m×1 m measurement light curtain composed of narrow-beam infrared emitter SFH 4550 and PIN photodiode arrays. It captured the photoelectric signals generated by the projectile’s obscuring beam to construct a complete algorithm framework of “signal acquisition - feature modeling - coordinate prediction”. The 8-bit binary sensor states were transformed into high-dimensional features through linear projection and positional encoding, and processed in the Transformer’s multi-head attention mechanism to effectively capture spatial relationships between different sensors. By acquiring and preprocessing signals with FPGA and performing feature modeling and coordinate prediction with the Transformer model, the system achieved accurate coordinate computation.Experimental results show that the proposed algorithm achieves high positioning accuracy in both single-point and multi-point occlusion scenarios, with the predicted coordinate error range controlled within ±1.5 mm. Compared with existing systems, the accuracy is significantly improved. Measurement verification based on high-precision coordinate target paper shows that the experimental accuracy meets practical application requirements, providing reliable technical support for the refinement and optimization of modern weapon systems. In addition, by comparing different models, the results validate that the Transformer-based feature extraction method in complex occlusion scenarios has good robustness and real-time performance.
LiuJi, SuNinggang, WuJinhui, et al. Development status of photographic detection technology for measuring projectile coordinates[J]. Journal of Sichuan Ordnance, 2018, 39(12): 224-230. (in Chinese)
XiaoYunfeng, YangDanlei, WangJinsong, et al. Light curtain target point coordinate measuring method[J]. Journal of Changchun University of Science and Technology, 2018, 41(2): 72-75. (in Chinese)
ChengRiwei, LanGuofeng, YanHaoming,et al.Design of temperature control device for laser light curtain velocity measurement system[J]. Journal of North University of China(Natural Science Edition), 2022, 43(4): 370-376. (in Chinese)
ZhangJianping, LiuJi, ZhangBin, et al. Projectile impact location measurement based on N shape laser screen[J]. Journal of Detection & Control, 2015, 37(1): 76-79. (in Chinese)
WangJia, ZhangZhixuan, DongTao, et al. Research on the measurement method of the vertical target of the six light screen based on double N-type[J]. Optical Technique, 2024, 50(2): 188-193. (in Chinese)
FengBin, ShiXiuhua, KangZhiqiang, et al. Study on measurement of impact point coordinates with the combination of acoustic array and optical detection[J]. Piezoelectrics & Acoustooptics, 2012, 34(2): 177-179. (in Chinese)
FengBin, WuZhichao. Two measuring algorithms for projectiles impact point coordinates based on sky screens vertical target[J]. Ordnance Industry Automation, 2013, 32(11): 63-66. (in Chinese)
ChenRui, NiJinping, HouSa. Three accuracy assessment methods of automatic projectile impact coordinates measurement system[J]. Journal of Test and Measurement Technology, 2017, 31(1): 69-75. (in Chinese)
[37]
TriyonoL, GernowoR, Prayitno.MoNetViT: an efficient fusion of CNN and transformer technologies for visual navigation assistance with multi query attention[J].Frontiers in Computer Science, 2025, 7: 1510252-1510252.
[38]
ChenD, DengC H, YangZ Y, et al. Accuracy calibration of the multi‑light screen measurement system using array optical signal[J]. Measurement Science and Technology, 2024, 35(9): 095008.
[39]
DongT, LiS, ChenD, et al. Measurement method for the flight parameters of a small flying object using a multi‑dimensional LED detection array[J]. Photonics, 2023, 10(12): 1313.
JinYufeng, TaoChongben. Fusion information enhanced method based on transformer for 3D object detection[J]. Chinese Journal of Scientific Instrument, 2023, 44(12): 297-306. (in Chinese)
MengXiangfu, ShiHaoyuan. Survey of transformer-based model for time series forecasting[J]. Journal of Frontiers of Computer Science & Technology, 2025, 19(1): 45-64. (in Chinese)
TangDonglin, YangZhou, ChengHeng, et al. Metal defect image recognition method based on shallow CNN fusion transformer[J]. China Mechanical Engineering, 2022, 33(19): 2298-2305. (in Chinese)