With the rapid advancement of autonomous driving, security surveillance, and virtual reality technologies, the demand for wide-field-of-view imaging systems continues to grow to achieve broader visual perception ranges. Inspired by the compound eyes of insects in nature, a bionic compound eye imaging system is designed based on a multi-camera array. Comprising eight cameras arranged in a spherical ring to mimic the natural compound eye structure, a wide field of view of 105°×100° is achieved in this system. To address common issues in wide-field image stitching, such as poor quality, ghosting, and chromatic aberration, a coarse-to-fine image stitching method guided by agglomerative hierarchical clustering, named AHC-LightGlue, is proposed. Firstly, key points and descriptors are extracted from the images using SuperPoint. Before feature matching, the topological structure of feature points is constructed by the agglomerative hierarchical clustering algorithm. Coarse matching is performed at the cluster level to lock corresponding regions, followed by fine matching within these regions using LightGlue to address high mismatch rates and misalignment issues. Subsequently, the homography matrix is calculated by marginalizing sample consensus. Finally, a natural transition between the reference and target images is obtained by the proposed adaptive weighting fusion method. In this approach, superior fusion results are delivered even under significant brightness differences, large-scale color discrepancies are reduced, and image stitching quality is further enhanced. Experiments demonstrate that both image quality and stitching accuracy are improved by the proposed algorithm and color discrepancies are effectively reduced in image stitching to form smoother and more natural stitched images. Compared to traditional methods, the image stitching method based on a multi-camera array bionic compound eye has great robustness and adaptability.
FuM, LiangH, ZhuC, et al. Image stitching techniques applied to plane or 3-D models: a review[J]. IEEE Sensors Journal, 2023, 23(8): 8060-8079.
[2]
QiQ, FuR, ShaoZ, et al. Multi-aperture optical imaging systems and their mathematical light field acquisition models[J]. Frontiers of Information Technology & Electronic Engineering, 2022, 23(6): 823-844.
[3]
TudelaR, BrücknerA, DuparréJ, et al. An image restoration approach for artificial compound eyes[J]. Image Processing: Algorithms and Systems VI, 2008, 6812: 68120O.
[4]
FloreanoD, Pericet-CamaraR, ViolletS, et al. Miniature curved artificial compound eyes[J]. Proceedings of the National Academy of Sciences of the United States of America, 2013, 110(23): 9267-9272.
[5]
AfshariH, JacquesL, BagnatoL, et al. The PANOPTIC camera: a plenoptic sensor with real-time omnidirectional capability[J]. Journal of Signal Processing Systems, 2013, 70(3): 305-328.
[6]
LiuS, WangJ, YuanR, et al. et al. Real-time and ultrahigh accuracy image synthesis algorithm for full field of view imaging system[J]. Scientific Reports, 2020, 10: 12389.
[7]
QiaoZ, HaoQ, LiuM, et al. Multicamera wide-field-of-view compound eye imaging system[C]//Optoelectronic Imaging and Multimedia Technology XI, 2024, 13239: 132391R.
[8]
ZhengY, ZhangH, LiX, et al. Fast-zoom and high-resolution sparse compound-eye camera based on dual-end collaborative optimization[J]. Opto-Electronic Advances, 2025, 8(6): 240285.
[9]
GeR, LiuD, ZhouH, et al. Real-time instance segmentation based on contour learning[J]. Journal of Measurement Science and Instrumentation, 2024, 15(3): 328-337.
GuoYanan, ChenYan, WangKangyi, et al. Lossless fundus image compression algorithm based on content adaptive blocking[J]. Journal of North University of China(Natural Science Edition), 2025, 46(4): 430-437. (in Chinese)
JiWenfang, ZhuZiwen, DengDezhi, et al. Rapid urban aerial image stitching method based on improved SIFT algorithm[J]. Journal of Test and Measurement Technology, 2024, 38(5): 500-505. (in Chinese)
[14]
LoweD G. Distinctive image features from scale-invariant keypoints[J]. International Journal of Computer Vision, 2004, 60(2): 91-110.
[15]
MujaM, LoweD G. Scalable nearest neighbor algorithms for high dimensional data[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(11): 2227-2240.
[16]
XiuC, FangJ, ZhangJ. Image stitching method based on adaptive weighted fusion[C]//2021 33rd Chinese Control and Decision Conference (CCDC), IEEE, 2021: 3099-3103.
[17]
DetoneD, MalisiewiczT, RabinovichA. SuperPoint: self-supervised interest point detection and description[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2018: 337-33712.
[18]
SunJ, ShenZ, WangY, et al. LoFTR: detector-free local feature matching with transformers[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021: 8918-8927.
[19]
SarlinP E, DetoneD, MalisiewiczT, et al. Superglue: learning feature matching with graph neural networks[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: 4937-4946.
[20]
LindenbergerP, SarlinP E, PollefeysM. Lightglue: local feature matching at light speed[C]//2023 IEEE/CVF International Conference on Computer Vision (ICCV), 2023: 17581-17592.
BalntasV, LencK, VedaldiA, et al. HPatches: a benchmark and evaluation of handcrafted and learned local descriptors[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 3852-3861.
[23]
LinK, JiangN, CheongL F, et al. SEAGULL: seam-guided local alignment for parallax-tolerant image stitching[C]//Computer Vision-ECCV 2016. Cham: Springer, 2016: 370-385.
[24]
DusmanuM, RoccoI, PajdlaT, et al. D2-net: a trainable CNN for joint description and detection of local features[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019: 8084-8093.
[25]
LiaoT, LiN. Single-perspective warps in natural image stitching[J]. IEEE Transactions on Image Processing, 2020, 29: 724-735.
[26]
NieL, LinC, LiaoK, et al. Parallax-tolerant unsupervised deep image stitching[C]//2023 IEEE/CVF International Conference on Computer Vision (ICCV), 2023: 7365-7374.