A New Compression Sensing Reconstruction Method for Stitching Images Based on Deformable Convolution-Deformable Deconvolution Module

A New Compression Sensing Reconstruction Method for Stitching Images Based on Deformable Convolution-Deformable Deconvolution Module | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article A New Compression Sensing Reconstruction Method for Stitching Images Based on Deformable Convolution-Deformable Deconvolution Module Jinwang Cha, Xing Hu, Li Xiao, Minqin Fan, Juan Xie, Cheng Pan, and 13 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7067314/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The compression sensing reconstruction of the stitching image is important for the communication of autonomous driving, intelligent vehicles, and unmanned aerial vehicles (UAV). An compression sensing reconstruction method for the stitching image is proposed. The proposed compression sensing reconstruction method contain: image stitching method, deformable convolution module, SCNet, and deformable deconvolution module. The added deformable convolution module can extract the features of the stitching image reasonably due to image distortion. The added deformable deconvolution module can make the stitching image and the reconstructed stitching image maintain consistency. Experimental results show the reconstruction SSIM and PSNR of the proposed method is better than the image stitching method and the SCNet. The reconstruction stitching images also has better visual effects. Physical sciences/Engineering Physical sciences/Mathematics and computing Physical sciences/Optics and photonics Figures Figure 1 Figure 2 1. Introduction The compression sensing reconstruction of stitching images is important for the communication of autonomous driving, intelligent vehicles, and unmanned aerial vehicles (UAV). The compression sensing reconstruction of the normal image has been well researched. The compression sensing reconstruction methods of the normal image can be divided two categories: 1) no-deep-learning-method and 2) deep-learning-method. To obtain the reconstruction results of larger field-of-view (FoV) images, the images need be stitched. The image stitching methods can also be devided into two categories: 1) no-deep-learning-method and 2) deep-learning-method. The stitching images can obtain larger field-of-view (FoV) compared with normal images. Therefore, the compresseion sensing reconstruction of stitching images can obtain larger reconstruction images compared with the compression sensing reconstruction of normal images. An compression sensing reconstruction method for the stitching image is proposed. The proposed compression sensing reconstruction method contain: image stitching method, deformable convolution module, SCNet, and deformable deconvolution module. The added deformable convolution module can extract the features of the stitching image reasonably due to image distortion. The added deformable deconvolution module can make the stitching image and the reconstructed stitching image maintain consistency. Experimental results show the reconstruction SSIM and PSNR of the proposed method is better than the image stitching method and the SCNet. The reconstruction stitching images also has better visual effects. 2. Related Work 2.1 The Compression Sensing Reconstruction Methods of the Normal Image 2.1.1 no-deep-learning-method At present, researchers in the field of compressive sensing have proposed many methods that can accurately reconstruct the original data, mainly divided into three categories: convex optimization methods[ 1 ], greedy methods[ 2 ], and combinatorial methods[ 3 ]. Among them, greedy methods have been widely used. Common greedy methods include: Orthogonal Matching Pursuit (OMP)[ 4 ], Regularized OMP (ROMP)[ 5 ], Compressive Sampling Matching Pursuit (CoSaMP)[ 6 ], Subspace Pursuit (SP)[ 7 ], Stagewise OMP (StOMP)[ 8 ], and Sparsity Adaptive Matching Pursuit (SAMP)[ 9 ], etc. 2.1.2 deep-learning-method Chen et al. [ 10 ] proposed a compression sesning reconstruction method combining SCL and SCNet. Yang and Yuan [ 11 ] designed a ultra-lightweight image compressive sensing reconstruction method based on knowledge distillation. Chen and Zhang [ 12 ] proposed a practical compact deep compressed sensing method. 2.2 The Image Stitching Methods of Normal Images 2.2.1 no-deep-learning-method The no-deep-learning-method for the image stitching mainly caontains the step: 1) the feature extraction, 2) feature matching, 3) homograph transformation, and 4) image blending. 2.2.2 deep-learning-method Zhu et al. [ 13 ] improved a novel panorama generative model for synthesizing realistic and sharp-looking panorama, which does not require a large number of labeled ground-truth data. Sumantri S. J. et al. [ 14 ] designed a learning-based approach the reconstructs the scene in 360°×180° from a sparse set of conventional images. Wu et al. [ 15 ] tackled the problem of synthesizing a ground-view panorama image conditioned on a top-view aerial image, which is a challenging problem in this domain. 3. A New Compression Sensing Reconstruction Method of the Stitching Image The image stitching method uses the method propsoed by [ 16 ], which is implemented in python.The proposed compression sensing reconstruction method contain: image stitching method, deformable convolution module, SCNet, and deformable deconvolution module. The added deformable convolution module can extract the features of the stitching image reasonably due to image distortion. The added deformable deconvolution module can make the stitching image and the reconstructed stitching image maintain consistency. The propsoed method is named DCM-DDM-SCNet. 4. Experimental Results 4.1 The Ablation Experiments The traning epoch of the SCNet is 200, the learning rate is 0.0001, the block size is 33, and the number of features is 32. Table I The Reconstruction Result SSIM (CS Ratio:0.01) PSNR(CS Ratio:0.01) SSIM(CS Ratio:0.04) PSNR(CS Ratio:0.04) IS + SCNet 58.35 14.17 62.41 16.43 IS + DCM + SCNet + DDM 60.45 15.39 63.26 18.58 The reconstruction SSIM and PSNR is shown in Table I. From Table I, the image stitching method + SCNet is named IS + SCNet, and the image stitching method + deformable convolution module + SCNet + deformable deconvolution module is named IS + DCM + SCNet + DDM. From Table I, in CS ratio 0.01, the SSIM and PSNR of IS + DCM + SCNet + DDM are better than the SSIM and PSNR of IS + SCNet. From Table I, in CS ratio 0.04, the SSIM and PSNR of IS + DCM + SCNet + DDM are also better than the SSIM and PSNR of IS + SCNet. 4.2 The Reconstruction Time Table II The Reconstruction Time FPS SCNet 11 DCM + SCNet + DDM 10 Table III The Image Stitching Time Image Stitching Time (s) Scene1 0.33 Scene2 0.34 The reconstruction time and the image stitching time are shown in Table II and Table III respectively. The FPS of SCNet is 11 and the FPS of DCM + SCNet + DDM is 10. The image stitching time of Scene1 is 0.33s and the image stitching time of Scene2 is 0.34s. Declarations Competing interests: The authors declare that there is no conflict of interest regarding the publication of this article. Author Contribution Xing Hu and Jinwang Cha contribute to the writing of this paper; Weihua Liu and Shuqin Wang contribute to the writing skills improvment of this paper; Li Xiao, Yi An, Cheng Shao, Jie Ren and Xinjian Li give much improvments on technologines; Ruifeng Pan, Hui Zhang, Mengsheng Wang, Jiapeng Zhu and Hongsheng Tian contribute to the experimental evaluation of this paper. Acknowledgement Thanks for my students Lin Wu, Xinyu Fan, Jiapeng Zhu, Yajuan Xiao, Yang Cao, and Yanchun Tian. Lin Wu, Xinyu Fan, and Shunan Zhao contribute to the writing skills improvement of this paper. Jiapeng Zhu and Yajuan Xiao contribute to the experimental evaluation of this paper. Yang Cao and Yanchun Tian give me much care in daily life. Data Availability: https://download.csdn.net/download/accdgh/90971115 References Candès, E. J., Romberg, J., & Tao, T. (2006). Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information. IEEE Transactions on information theory, 52(2), 489–509. Needell, D., & Vershynin, R. (2010). Signal recovery from incomplete and inaccurate measurements via regularized orthogonal matching pursuit. IEEE Journal of selected topics in signal processing, 4(2), 310–316. Khan, I., & Singh, D. (2018). Efficient compressive sensing based sparse channel estimation for 5G massive MIMO systems. AEU-International Journal of Electronics and Communications, 89, 181–190. Tropp, J. A., & Gilbert, A. C. (2007). Signal recovery from random measurements via orthogonal matching pursuit. IEEE Transactions on information theory, 53(12), 4655–4666. Yang, M., & De Hoog, F. (2015). Orthogonal matching pursuit with thresholding and its application in compressive sensing. IEEE Transactions on Signal Processing, 63(20), 5479–5486. Needell, D., & Tropp, J. A. (2009). CoSaMP: Iterative signal recovery from incomplete and inaccurate samples. Applied and computational harmonic analysis, 26(3), 301–321. Dai, W., & Milenkovic, O. (2009). Subspace pursuit for compressive sensing signal reconstruction. IEEE transactions on Information Theory, 55(5), 2230–2249. Marques, E. C., Maciel, N., Naviner, L., Cai, H., & Yang, J. (2018). A review of sparse recovery algorithms. IEEE access, 7, 1300–1322. Ba, K. D., Indyk, P., Price, E., & Woodruff, D. P. (2010, January). Lower bounds for sparse recovery. In Proceedings of the twenty-first annual ACM-SIAM symposium on Discrete Algorithms (pp. 1190–1197). Society for Industrial and Applied Mathematics. Chen, Bin, et al. "Self-supervised scalable deep compressed sensing." International Journal of Computer Vision 133.2 (2025): 688–723. Yang, Y., & Yuan, W. (2023, July). Ultra-lightweight Image Compressive Sensing Reconstruction Algorithm Based on Knowledge Distillation. In 2023 3rd International Symposium on Computer Technology and Information Science (ISCTIS) (pp. 232–237). IEEE. Chen, B., & Zhang, J. (2024). Practical compact deep compressed sensing. IEEE Transactions on Pattern Analysis and Machine Intelligence. Zhu D., Zhou Q., Han T., and Chen Y., "360 Degree Panorama Synthesis from Sequential Views Based on Improved FC-densenets", IEEE Access, 2019, 7: 180503–180511, https://doi.org/10.1109/ACCESS.2019.2958111 Sumantri J. S. and Park I. K., "360 Panorama Synthesis from a Sparse Set of Images on a Low-Power Device", IEEE Trans. Comput. Imag., 2020, 6: 1179–1193, https://doi.org/10.1109/TCI.2020.3011854 Wu S., Tang H., Jing Xiao-Y., Zhao H., Qian J., Sebe N., and Yan Y., "Cross-View Panorama Image Synthesis", IEEE Trans. Multimedia, 2023, 25: 3546–3559, https://doi.org/10.1109/TMM.2022.3162474 Ribeiro D., CustÓdio P., and Balasubramaniam L., "Image Stitching and 3D Point Cloud Registration", Image Processing and Vision MEEC, 2021. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7067314","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":481900393,"identity":"148b8d6e-5c82-4f05-9b60-fce48caaa576","order_by":0,"name":"Jinwang Cha","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Jinwang","middleName":"","lastName":"Cha","suffix":""},{"id":481900399,"identity":"a0c2848e-36e9-4af9-ac95-16ab90d6e61f","order_by":1,"name":"Xing Hu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIiWNgGAWjYBACfvbGBoPEP//k+NkbiNQi2XO4oeBhwwFjyZ4DRGoxuOHe8BGoJXHDjQSitTA2bkjccSdx5szHG28w1NhEE3bY7cZmg8Qzz4z7pdOKLRiOpeU2ENLCd+dgm0ECG7PszNk5ZhKMDYcJa2G4kdj+A6iFccPNM0RqEbiRCAzktsOKG27wEKlFsudgg0HCmTRgIAP9kkCMX/jZ2x8Y/qiwAUbl4Y03PtTYEOEXJGAgkUCKcogWUnWMglEwCkbByAAAxs1LURAEGdkAAAAASUVORK5CYII=","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":true,"prefix":"","firstName":"Xing","middleName":"","lastName":"Hu","suffix":""},{"id":481900402,"identity":"8af2cf50-7b1b-4cc9-91c5-e9e564700a9a","order_by":2,"name":"Li Xiao","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Xiao","suffix":""},{"id":481900406,"identity":"f14c1356-632c-4a92-ab5f-d374882d8efc","order_by":3,"name":"Minqin Fan","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Minqin","middleName":"","lastName":"Fan","suffix":""},{"id":481900408,"identity":"051679b6-59a8-4a4e-a75d-7c4e432511a5","order_by":4,"name":"Juan Xie","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Juan","middleName":"","lastName":"Xie","suffix":""},{"id":481900412,"identity":"8d05734e-57da-45b1-8277-58434a66dbce","order_by":5,"name":"Cheng Pan","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Cheng","middleName":"","lastName":"Pan","suffix":""},{"id":481900417,"identity":"1b650bdf-db97-448c-aa73-8facd88e839c","order_by":6,"name":"Jian Zheng","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Jian","middleName":"","lastName":"Zheng","suffix":""},{"id":481900420,"identity":"33035cb2-5167-454d-bf2d-fc55c52007c3","order_by":7,"name":"Ruifeng Pan","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Ruifeng","middleName":"","lastName":"Pan","suffix":""},{"id":481900424,"identity":"77c25388-6341-475c-8ab9-98701200f4ea","order_by":8,"name":"Yi An","email":"","orcid":"","institution":"Dalian University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"An","suffix":""},{"id":481900427,"identity":"3e001334-f9f3-4ffd-b50f-6f8367ed111d","order_by":9,"name":"Cheng Shao","email":"","orcid":"","institution":"Dalian University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Cheng","middleName":"","lastName":"Shao","suffix":""},{"id":481900429,"identity":"ea9fe266-4072-4beb-9238-665c2bfd4ac4","order_by":10,"name":"Jie Ren","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Jie","middleName":"","lastName":"Ren","suffix":""},{"id":481900435,"identity":"1aa35966-8690-45de-a62c-e1976fec433c","order_by":11,"name":"Xinjian Li","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Xinjian","middleName":"","lastName":"Li","suffix":""},{"id":481900437,"identity":"0052439b-1fe3-4fc1-9320-62be34031ef1","order_by":12,"name":"Hui Zhang","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Hui","middleName":"","lastName":"Zhang","suffix":""},{"id":481900438,"identity":"6a4ba686-f042-4c9e-babc-d7a31408c217","order_by":13,"name":"Mengsheng Wang","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Mengsheng","middleName":"","lastName":"Wang","suffix":""},{"id":481900439,"identity":"326dbe3c-acbf-4861-8aa6-ee09c157b98f","order_by":14,"name":"Jiapeng Zhu","email":"","orcid":"","institution":"Nanchang Vocational University","correspondingAuthor":false,"prefix":"","firstName":"Jiapeng","middleName":"","lastName":"Zhu","suffix":""},{"id":481900440,"identity":"9772020c-23e4-4b85-b8f1-09f7d50d757e","order_by":15,"name":"Hongsheng Tian","email":"","orcid":"","institution":"State Grid Tieling Electric Power Supply Company","correspondingAuthor":false,"prefix":"","firstName":"Hongsheng","middleName":"","lastName":"Tian","suffix":""},{"id":481900441,"identity":"a52b5924-35ee-4099-8fa2-d191e2c0c170","order_by":16,"name":"Qingru Guo","email":"","orcid":"","institution":"Product Development Department, Googosoft","correspondingAuthor":false,"prefix":"","firstName":"Qingru","middleName":"","lastName":"Guo","suffix":""},{"id":481900442,"identity":"9a35942a-99b2-4e24-a178-6eb70a69428b","order_by":17,"name":"Wen Xiong","email":"","orcid":"","institution":"CCCC Mechanical and Electrical Engineering Bureau Co.,Ltd.Wuhan Design Institute Branch","correspondingAuthor":false,"prefix":"","firstName":"Wen","middleName":"","lastName":"Xiong","suffix":""},{"id":481900443,"identity":"9722d795-7107-4afb-b217-7feec3f57ba6","order_by":18,"name":"Shiming Chen","email":"","orcid":"","institution":"Weichai Heavy Machinery Co., Ltd","correspondingAuthor":false,"prefix":"","firstName":"Shiming","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2025-07-07 16:13:11","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-7067314/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7067314/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":86422332,"identity":"c5a1ec10-56fa-407a-aab1-113114c16e58","added_by":"auto","created_at":"2025-07-10 12:57:30","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":143566,"visible":true,"origin":"","legend":"\u003cp\u003eThe Reconstruction Image Scene 1\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7067314/v1/3d8cb1d9aa5d42d345d45571.png"},{"id":86421193,"identity":"8ed68df9-a5f9-4344-9f6e-9d3952b815ee","added_by":"auto","created_at":"2025-07-10 12:49:30","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":106352,"visible":true,"origin":"","legend":"\u003cp\u003eThe Reconstruction Image Scene 2\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7067314/v1/fb359e7e00c15cbf90d17a10.png"},{"id":87235967,"identity":"79d5a2ac-dfe0-4825-b5d7-91f8c3195bd3","added_by":"auto","created_at":"2025-07-21 21:46:25","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":941277,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7067314/v1/6ae074c8-adf7-4cc3-aee8-c38cb9ca675c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A New Compression Sensing Reconstruction Method for Stitching Images Based on Deformable Convolution-Deformable Deconvolution Module","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eThe compression sensing reconstruction of stitching images is important for the communication of autonomous driving, intelligent vehicles, and unmanned aerial vehicles (UAV). The compression sensing reconstruction of the normal image has been well researched. The compression sensing reconstruction methods of the normal image can be divided two categories: 1) no-deep-learning-method and 2) deep-learning-method.\u003c/p\u003e\u003cp\u003eTo obtain the reconstruction results of larger field-of-view (FoV) images, the images need be stitched. The image stitching methods can also be devided into two categories: 1) no-deep-learning-method and 2) deep-learning-method.\u003c/p\u003e\u003cp\u003eThe stitching images can obtain larger field-of-view (FoV) compared with normal images. Therefore, the compresseion sensing reconstruction of stitching images can obtain larger reconstruction images compared with the compression sensing reconstruction of normal images. An compression sensing reconstruction method for the stitching image is proposed. The proposed compression sensing reconstruction method contain: image stitching method, deformable convolution module, SCNet, and deformable deconvolution module. The added deformable convolution module can extract the features of the stitching image reasonably due to image distortion. The added deformable deconvolution module can make the stitching image and the reconstructed stitching image maintain consistency. Experimental results show the reconstruction SSIM and PSNR of the proposed method is better than the image stitching method and the SCNet. The reconstruction stitching images also has better visual effects.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"2. Related Work","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 The Compression Sensing Reconstruction Methods of the Normal Image\u003c/h2\u003e\u003cdiv id=\"Sec4\" class=\"Section3\"\u003e\u003ch2\u003e2.1.1 no-deep-learning-method\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eAt present, researchers in the field of compressive sensing have proposed many methods that can accurately reconstruct the original data, mainly divided into three categories: convex optimization methods[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], greedy methods[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e], and combinatorial methods[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Among them, greedy methods have been widely used. Common greedy methods include: Orthogonal Matching Pursuit (OMP)[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], Regularized OMP (ROMP)[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], Compressive Sampling Matching Pursuit (CoSaMP)[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e], Subspace Pursuit (SP)[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e], Stagewise OMP (StOMP)[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], and Sparsity Adaptive Matching Pursuit (SAMP)[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], etc.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section3\"\u003e\u003ch2\u003e2.1.2 deep-learning-method\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eChen et al. [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] proposed a compression sesning reconstruction method combining SCL and SCNet. Yang and Yuan [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] designed a ultra-lightweight image compressive sensing reconstruction method based on knowledge distillation. Chen and Zhang [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e] proposed a practical compact deep compressed sensing method.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.2 The Image Stitching Methods of Normal Images\u003c/h2\u003e\u003cdiv id=\"Sec7\" class=\"Section3\"\u003e\u003ch2\u003e2.2.1 no-deep-learning-method\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eThe no-deep-learning-method for the image stitching mainly caontains the step: 1) the feature extraction, 2) feature matching, 3) homograph transformation, and 4) image blending.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section3\"\u003e\u003ch2\u003e2.2.2 deep-learning-method\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eZhu et al. [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] improved a novel panorama generative model for synthesizing realistic and sharp-looking panorama, which does not require a large number of labeled ground-truth data. Sumantri S. J. et al. [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] designed a learning-based approach the reconstructs the scene in 360\u0026deg;\u0026times;180\u0026deg; from a sparse set of conventional images. Wu et al. [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] tackled the problem of synthesizing a ground-view panorama image conditioned on a top-view aerial image, which is a challenging problem in this domain.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"3. A New Compression Sensing Reconstruction Method of the Stitching Image","content":"\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eThe image stitching method uses the method propsoed by [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], which is implemented in python.The proposed compression sensing reconstruction method contain: image stitching method, deformable convolution module, SCNet, and deformable deconvolution module. The added deformable convolution module can extract the features of the stitching image reasonably due to image distortion. The added deformable deconvolution module can make the stitching image and the reconstructed stitching image maintain consistency. The propsoed method is named DCM-DDM-SCNet.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"4. Experimental Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003e4.1 The Ablation Experiments\u003c/h2\u003e\n \u003cdiv class=\"BlockQuote\"\u003e\n \u003cp\u003eThe traning epoch of the SCNet is 200, the learning rate is 0.0001, the block size is 33, and the number of features is 32.\u003c/p\u003e\n \u003cp\u003eTable I The Reconstruction Result\u003c/p\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Taba\" border=\"1\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSSIM (CS Ratio:0.01)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePSNR(CS Ratio:0.01)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSSIM(CS Ratio:0.04)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePSNR(CS Ratio:0.04)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIS\u0026thinsp;+\u0026thinsp;SCNet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e58.35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e62.41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16.43\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIS\u0026thinsp;+\u0026thinsp;DCM\u0026thinsp;+\u0026thinsp;SCNet\u0026thinsp;+\u0026thinsp;DDM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e60.45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15.39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e63.26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e18.58\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cdiv class=\"BlockQuote\"\u003e\n \u003cp\u003eThe reconstruction SSIM and PSNR is shown in Table I. From Table I, the image stitching method\u0026thinsp;+\u0026thinsp;SCNet is named IS\u0026thinsp;+\u0026thinsp;SCNet, and the image stitching method\u0026thinsp;+\u0026thinsp;deformable convolution module\u0026thinsp;+\u0026thinsp;SCNet\u0026thinsp;+\u0026thinsp;deformable deconvolution module is named IS\u0026thinsp;+\u0026thinsp;DCM\u0026thinsp;+\u0026thinsp;SCNet\u0026thinsp;+\u0026thinsp;DDM. From Table I, in CS ratio 0.01, the SSIM and PSNR of IS\u0026thinsp;+\u0026thinsp;DCM\u0026thinsp;+\u0026thinsp;SCNet\u0026thinsp;+\u0026thinsp;DDM are better than the SSIM and PSNR of IS\u0026thinsp;+\u0026thinsp;SCNet. From Table I, in CS ratio 0.04, the SSIM and PSNR of IS\u0026thinsp;+\u0026thinsp;DCM\u0026thinsp;+\u0026thinsp;SCNet\u0026thinsp;+\u0026thinsp;DDM are also better than the SSIM and PSNR of IS\u0026thinsp;+\u0026thinsp;SCNet.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003e4.2 The Reconstruction Time\u003c/h2\u003e\n \u003cdiv class=\"BlockQuote\"\u003e\n \u003cp\u003eTable II The Reconstruction Time\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"char\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tabb\" border=\"1\"\u003e\n \u003cthead\u003e\n \u003ctr style=\"height: 35px;\"\u003e\n \u003cth style=\"height: 35px;\" align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth style=\"height: 35px;\" align=\"left\"\u003e\n \u003cp\u003eFPS\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr style=\"height: 35px;\"\u003e\n \u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n \u003cp\u003eSCNet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"height: 35px;\" align=\"char\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr style=\"height: 35px;\"\u003e\n \u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n \u003cp\u003eDCM\u0026thinsp;+\u0026thinsp;SCNet\u0026thinsp;+\u0026thinsp;DDM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"height: 35px;\" align=\"char\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"char\"\u003e\n \u003cp\u003eTable III The Image Stitching Time\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"char\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tabc\" border=\"1\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eImage Stitching Time (s)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eScene1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.33\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eScene2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eThe reconstruction time and the image stitching time are shown in Table II and Table III respectively. The FPS of SCNet is 11 and the FPS of DCM\u0026thinsp;+\u0026thinsp;SCNet\u0026thinsp;+\u0026thinsp;DDM is 10. The image stitching time of Scene1 is 0.33s and the image stitching time of Scene2 is 0.34s.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCompeting interests:\u003c/h2\u003e\u003cp\u003eThe authors declare that there is no conflict of interest regarding the publication of this article.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eXing Hu and Jinwang Cha contribute to the writing of this paper; Weihua Liu and Shuqin Wang contribute to the writing skills improvment of this paper; Li Xiao, Yi An, Cheng Shao, Jie Ren and Xinjian Li give much improvments on technologines; Ruifeng Pan, Hui Zhang, Mengsheng Wang, Jiapeng Zhu and Hongsheng Tian contribute to the experimental evaluation of this paper.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThanks for my students Lin Wu, Xinyu Fan, Jiapeng Zhu, Yajuan Xiao, Yang Cao, and Yanchun Tian. Lin Wu, Xinyu Fan, and Shunan Zhao contribute to the writing skills improvement of this paper. Jiapeng Zhu and Yajuan Xiao contribute to the experimental evaluation of this paper. Yang Cao and Yanchun Tian give me much care in daily life.\u003c/p\u003e\u003ch2\u003eData Availability:\u003c/h2\u003e\u003cp\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://download.csdn.net/download/accdgh/90971115\u003c/span\u003e\u003cspan address=\"https://download.csdn.net/download/accdgh/90971115\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eCand\u0026egrave;s, E. J., Romberg, J., \u0026amp; Tao, T. (2006). Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information. IEEE Transactions on information theory, 52(2), 489\u0026ndash;509.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNeedell, D., \u0026amp; Vershynin, R. (2010). Signal recovery from incomplete and inaccurate measurements via regularized orthogonal matching pursuit. IEEE Journal of selected topics in signal processing, 4(2), 310\u0026ndash;316.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKhan, I., \u0026amp; Singh, D. (2018). Efficient compressive sensing based sparse channel estimation for 5G massive MIMO systems. AEU-International Journal of Electronics and Communications, 89, 181\u0026ndash;190.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTropp, J. A., \u0026amp; Gilbert, A. C. (2007). Signal recovery from random measurements via orthogonal matching pursuit. IEEE Transactions on information theory, 53(12), 4655\u0026ndash;4666.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang, M., \u0026amp; De Hoog, F. (2015). Orthogonal matching pursuit with thresholding and its application in compressive sensing. IEEE Transactions on Signal Processing, 63(20), 5479\u0026ndash;5486.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNeedell, D., \u0026amp; Tropp, J. A. (2009). CoSaMP: Iterative signal recovery from incomplete and inaccurate samples. Applied and computational harmonic analysis, 26(3), 301\u0026ndash;321.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDai, W., \u0026amp; Milenkovic, O. (2009). Subspace pursuit for compressive sensing signal reconstruction. IEEE transactions on Information Theory, 55(5), 2230\u0026ndash;2249.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMarques, E. C., Maciel, N., Naviner, L., Cai, H., \u0026amp; Yang, J. (2018). A review of sparse recovery algorithms. IEEE access, 7, 1300\u0026ndash;1322.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBa, K. D., Indyk, P., Price, E., \u0026amp; Woodruff, D. P. (2010, January). Lower bounds for sparse recovery. In Proceedings of the twenty-first annual ACM-SIAM symposium on Discrete Algorithms (pp. 1190\u0026ndash;1197). Society for Industrial and Applied Mathematics.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen, Bin, et al. \"Self-supervised scalable deep compressed sensing.\" International Journal of Computer Vision 133.2 (2025): 688\u0026ndash;723.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang, Y., \u0026amp; Yuan, W. (2023, July). Ultra-lightweight Image Compressive Sensing Reconstruction Algorithm Based on Knowledge Distillation. In 2023 3rd International Symposium on Computer Technology and Information Science (ISCTIS) (pp. 232\u0026ndash;237). IEEE.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen, B., \u0026amp; Zhang, J. (2024). Practical compact deep compressed sensing. IEEE Transactions on Pattern Analysis and Machine Intelligence.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhu D., Zhou Q., Han T., and Chen Y., \"360 Degree Panorama Synthesis from Sequential Views Based on Improved FC-densenets\", IEEE Access, 2019, 7: 180503\u0026ndash;180511, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2019.2958111\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2019.2958111\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSumantri J. S. and Park I. K., \"360 Panorama Synthesis from a Sparse Set of Images on a Low-Power Device\", IEEE Trans. Comput. Imag., 2020, 6: 1179\u0026ndash;1193, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/TCI.2020.3011854\u003c/span\u003e\u003cspan address=\"10.1109/TCI.2020.3011854\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu S., Tang H., Jing Xiao-Y., Zhao H., Qian J., Sebe N., and Yan Y., \"Cross-View Panorama Image Synthesis\", IEEE Trans. Multimedia, 2023, 25: 3546\u0026ndash;3559, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/TMM.2022.3162474\u003c/span\u003e\u003cspan address=\"10.1109/TMM.2022.3162474\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRibeiro D., Cust\u0026Oacute;dio P., and Balasubramaniam L., \"Image Stitching and 3D Point Cloud Registration\", Image Processing and Vision MEEC, 2021.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7067314/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7067314/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe compression sensing reconstruction of the stitching image is important for the communication of autonomous driving, intelligent vehicles, and unmanned aerial vehicles (UAV). An compression sensing reconstruction method for the stitching image is proposed. The proposed compression sensing reconstruction method contain: image stitching method, deformable convolution module, SCNet, and deformable deconvolution module. The added deformable convolution module can extract the features of the stitching image reasonably due to image distortion. The added deformable deconvolution module can make the stitching image and the reconstructed stitching image maintain consistency. Experimental results show the reconstruction SSIM and PSNR of the proposed method is better than the image stitching method and the SCNet. The reconstruction stitching images also has better visual effects.\u003c/p\u003e","manuscriptTitle":"A New Compression Sensing Reconstruction Method for Stitching Images Based on Deformable Convolution-Deformable Deconvolution Module","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-10 12:49:26","doi":"10.21203/rs.3.rs-7067314/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"cdddb244-dbc5-4212-b7be-16e5ddf83689","owner":[],"postedDate":"July 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":51165765,"name":"Physical sciences/Engineering"},{"id":51165766,"name":"Physical sciences/Mathematics and computing"},{"id":51165767,"name":"Physical sciences/Optics and photonics"}],"tags":[],"updatedAt":"2025-07-21T21:38:18+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-10 12:49:26","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7067314","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7067314","identity":"rs-7067314","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}