研究生: |
鄭維新 Wei-Hsin Cheng |
---|---|
論文名稱: |
物聯網邊緣攝影機上多色彩空間特徵融合網路之空拍影像動態除霧 Dynamic Aerial Image Dehazing with Multi-color Spaces Feature Fusion Network for IoT-enabled Edge Cameras |
指導教授: |
陸敬互
Ching-Hu Lu |
口試委員: |
鍾聖倫
Sheng-Luen Chung 馬尚彬 Shang-Pin Ma 許嘉裕 Chia-Yu Hsu 廖峻鋒 Chun-Feng Liao |
學位類別: |
碩士 Master |
系所名稱: |
電資學院 - 電機工程系 Department of Electrical Engineering |
論文出版年: | 2023 |
畢業學年度: | 111 |
語文別: | 中文 |
論文頁數: | 93 |
中文關鍵詞: | 空拍影像除霧 、多色彩空間特徵融合 、輕量化深度網路 、動態除霧 、邊緣運算 、物聯網 |
外文關鍵詞: | aerial image dehazing, multi-color space feature fusion, lightweight deep network, dynamic dehazing, edge computing, IoT |
相關次數: | 點閱:336 下載:2 |
分享至: |
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
隨著物聯網 (IoT) 的發展,結合人工智慧的邊緣計算攝影機 (以下簡稱邊緣攝影機) 已能夠直接在邊緣端進行影像強化。近年來,已有研究採用深度神經網路進行空拍影像除霧。然而,既有的研究大多僅使用RGB色彩空間的影像進行除霧,或是需要額外使用其他影像來源進行除霧,未考慮不同的色彩空間中蘊含的特徵,導致無法充分利用既有的輸入。為了獲取更豐富的特徵,本研究首度提出「輕量化多色彩空間特徵融合空拍影像除霧網路」,將輸入影像透過色彩空間轉換來額外產生HSV以及YCbCr色彩空間,並使用特徵注意力模組將不同色彩空間的特徵進行融合並給予加權,使模型更加關注在重要的特徵。同時,使用了邊線增強模組,使特徵在模型裡傳遞的時候能最大化的保留邊線的資訊。實驗結果顯示,相較於最新研究,在Sate1K資料集中,未輕量化的網路平均使PSNR提升1.57%,SSIM提升3.48%,而輕量化的網路平均使PSNR提升0.11%,SSIM提升3.25%。在RICE資料集中,未輕量化的網路使PSNR提升5.00%,SSIM提升0.78%,輕量化的網路使PSNR提升3.49%,SSIM提升0.71%,因此可以證明本研究提出之網路可獲得品質更好的空拍除霧影像。接著,為了充分利用攝影機的運算資源,且避免對無霧的空拍影像進行不必要的處理,本研究加入「輕量化動態空拍影像除霧偵測網路」,該模型用於評估空拍影像中是否存在霧,以決定是否需要除霧。實驗結果顯示,將輕量化動態影像除霧偵測網路與輕量化多色彩空間特徵融合空拍影像除霧網路整合後,在空拍影像有霧的占比為0.2時,能使FPS提升218.90%、平均運行時間降低68.35%、PSNR提升196.15%以及SSIM提升4.69%。當占比為0.4時,能使FPS提升92.47%、平均運行時間降低47.44%、PSNR提升141.20%以及SSIM提升3.40%。當占比為0.6時,能使FPS提升35.24%、平均運行時間降低25.97%、PSNR提升91.11%以及SSIM提升2.27%。當占比為0.8時,能使FPS提升4.43%、平均運行時間降低5.13%、PSNR提升44.73%以及SSIM提升1.11%。因此可以證實動態評估影像可提高邊緣攝影機的運行效率以及輸出的影像品質。
With the development of the Internet of Things (IoT), smart cameras combined with artificial intelligence (AI) and edge computing technologies (hereafter referred to edge cameras) improve image quality directly on edge cameras. In recent years, deep neural networks have been employed for aerial image dehazing. However, existing research mainly focused on using RGB images or required additional images, neglecting the potential features in different color spaces and thus failing to fully utilize the input images. To address this, a novel "lightweight multi-color space feature fusion aerial image dehazing network" is proposed in this study. The input image is transformed into HSV and YCbCr color spaces, and a feature attention block is utilized to fuse and weight the features from different color spaces, enabling the model to focus more on crucial features. Additionally, a borderline boosting block is utilized to maximize the preservation of borderline details during feature propagation within the model. Experimental results demonstrate improvements over the latest research. In the Sate1K dataset, the non-lightweight network increases peak signal-to-noise ratio (PSNR) by 1.57% and structural similarity index (SSIM) by 3.48%, while the lightweight network increases PSNR by 0.11% and SSIM by 3.25%. On the RICE dataset, the non-lightweight network increases PSNR by 5.00% and SSIM by 0.78%, while the lightweight network increases PSNR by 3.49% and SSIM by 0.71%. These results validate that the proposed network can improve the quality of the dehazed aerial images. Furthermore, in order to effectively utilize the computational resources of edge cameras, this study introduces a "lightweight dynamic aerial image dehazing detector". This model first detects haze in aerial images to determine if dehazing is needed. Experimental results show that integrating this detection model with the dehazing network significantly improves performance. When the proportion of haze aerial images is 0.2, it increases the frames per second (FPS) by 218.90%, reduces the average processing time by 68.35%, increases PSNR by 196.15% and increases SSIM by 4.69%. At a proportion of 0.4, it increases FPS by 92.47%, reduces average processing time by 47.44%, increases PSNR by 141.20% and increases SSIM by 3.40%. At a proportion of 0.6, it increases FPS by 35.24%, reduces average processing time by 25.97%, increases PSNR by 91.11% and increases SSIM by 2.27%. At a proportion of 0.8, it increases FPS by 4.43%, reduces average processing time by 5.13%, increases PSNR by 44.73% and increases SSIM by 1.11%. This confirms that dynamically evaluating the image condition can improve the operational efficiency of edge cameras and the quality of output images.
[1] X. You, C. Zhang, X. Tan, S. Jin, and H. Wu, "AI for 5G: research directions and paradigms," Science China Information Sciences, vol. 62, pp. 1-13, 2019.
[2] I. Chiuchisan, I. Chiuchisan, and M. Dimian, "Internet of Things for e-Health: An approach to medical applications," in 2015 International Workshop on Computational Intelligence for Multimedia Understanding (IWCIM), 2015: IEEE, pp. 1-5.
[3] M. M. Rathore, A. Paul, A. Ahmad, and G. Jeon, "IoT-based big data: from smart city towards next generation super city planning," International Journal on Semantic Web and Information Systems (IJSWIS), vol. 13, no. 1, pp. 28-47, 2017.
[4] M. A. Rahim, M. A. Rahman, M. M. Rahman, A. T. Asyhari, M. Z. A. Bhuiyan, and D. Ramasamy, "Evolution of IoT-enabled connectivity and applications in automotive industry: A review," Vehicular Communications, vol. 27, p. 100285, 2021.
[5] D. McLean. "13 predictions about the trends that will shape smart cities in 2023." smartcities dive. https://www.smartcitiesdive.com/news/smart-cities-predictions-2023-curb-management-autonomous-vehicles-electric-vehicles-IoT-AI-ebikes/639004/ (accessed.
[6] K. Cao, Y. Liu, G. Meng, and Q. Sun, "An overview on edge computing research," IEEE access, vol. 8, pp. 85714-85728, 2020.
[7] H. Li, K. Ota, and M. Dong, "Learning IoT in edge: Deep learning for the Internet of Things with edge computing," IEEE network, vol. 32, no. 1, pp. 96-101, 2018.
[8] W. Yu et al., "A survey on the edge computing for the Internet of Things," IEEE access, vol. 6, pp. 6900-6919, 2017.
[9] Y. Wan, X. Hu, Y. Zhong, A. Ma, L. Wei, and L. Zhang, "Tailings reservoir disaster and environmental monitoring using the UAV-ground hyperspectral joint observation and processing: a case of study in Xinjiang, the belt and road," in IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium, 2019: IEEE, pp. 9713-9716.
[10] D. J. Mulla, "Twenty five years of remote sensing in precision agriculture: Key advances and remaining knowledge gaps," Biosystems engineering, vol. 114, no. 4, pp. 358-371, 2013.
[11] C. Holness, T. Matthews, K. Satchell, and E. C. Swindell, "Remote sensing archeological sites through unmanned aerial vehicle (UAV) imaging," in 2016 IEEE international geoscience and remote sensing symposium (IGARSS), 2016: IEEE, pp. 6695-6698.
[12] B. Kellenberger, M. Volpi, and D. Tuia, "Fast animal detection in UAV images using convolutional neural networks," in 2017 IEEE international geoscience and remote sensing symposium (IGARSS), 2017: IEEE, pp. 866-869.
[13] A. Kulkarni and S. Murala, "Aerial Image Dehazing With Attentive Deformable Transformers," in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2023, pp. 6305-6314.
[14] M. H. Asmare, V. S. Asirvadam, and L. Iznita, "Color space selection for color image enhancement applications," in 2009 International Conference on Signal Acquisition and Processing, 2009: IEEE, pp. 208-212.
[15] C.-H. Lu and B.-E. Shao, "Environment-aware multiscene image enhancement for internet of things enabled edge cameras," IEEE Systems Journal, vol. 15, no. 3, pp. 3439-3449, 2020.
[16] E. J. McCartney, "Optics of the atmosphere: scattering by molecules and particles," New York, 1976.
[17] J. Long, Z. Shi, W. Tang, and C. Zhang, "Single remote sensing image dehazing," IEEE Geoscience and Remote Sensing Letters, vol. 11, no. 1, pp. 59-63, 2013.
[18] K. He, J. Sun, and X. Tang, "Single image haze removal using dark channel prior," IEEE transactions on pattern analysis and machine intelligence, vol. 33, no. 12, pp. 2341-2353, 2010.
[19] X. Pan, F. Xie, Z. Jiang, and J. Yin, "Haze removal for a single remote sensing image based on deformed haze imaging model," IEEE Signal Processing Letters, vol. 22, no. 10, pp. 1806-1810, 2015.
[20] J. Li, Q. Hu, and M. Ai, "Haze and thin cloud removal via sphere model improved dark channel prior," IEEE Geoscience and Remote Sensing Letters, vol. 16, no. 3, pp. 472-476, 2018.
[21] S. Huang, D. Li, W. Zhao, and Y. Liu, "Haze removal algorithm for optical remote sensing image based on multi-scale model and histogram characteristic," IEEE Access, vol. 7, pp. 104179-104196, 2019.
[22] D. J. Jobson, Z.-u. Rahman, and G. A. Woodell, "Properties and performance of a center/surround retinex," IEEE transactions on image processing, vol. 6, no. 3, pp. 451-462, 1997.
[23] D. J. Jobson, Z.-u. Rahman, and G. A. Woodell, "A multiscale retinex for bridging the gap between color images and the human observation of scenes," IEEE Transactions on Image processing, vol. 6, no. 7, pp. 965-976, 1997.
[24] P. S. Chavez Jr, "An improved dark-object subtraction technique for atmospheric scattering correction of multispectral data," Remote sensing of environment, vol. 24, no. 3, pp. 459-479, 1988.
[25] A. Makarau, R. Richter, R. Müller, and P. Reinartz, "Haze detection and removal in remotely sensed multispectral imagery," IEEE Transactions on Geoscience and Remote Sensing, vol. 52, no. 9, pp. 5895-5905, 2014.
[26] Q. Liu, X. Gao, L. He, and W. Lu, "Haze removal for a single visible remote sensing image," Signal Processing, vol. 137, pp. 33-43, 2017.
[27] B. Huang, Y. Li, X. Han, Y. Cui, W. Li, and R. Li, "Cloud removal from optical satellite imagery with SAR imagery using sparse representation," IEEE Geoscience and Remote Sensing Letters, vol. 12, no. 5, pp. 1046-1050, 2015.
[28] H. Pan, "Cloud removal for remote sensing imagery via spatial attention generative adversarial network," arXiv preprint arXiv:2009.13015, 2020.
[29] I. Goodfellow et al., "Generative adversarial networks," Communications of the ACM, vol. 63, no. 11, pp. 139-144, 2020.
[30] K. Enomoto et al., "Filmy cloud removal on satellite imagery with multispectral conditional generative adversarial nets," in Proceedings of the IEEE conference on computer vision and pattern recognition workshops, 2017, pp. 48-56.
[31] M. Mirza and S. Osindero, "Conditional generative adversarial nets," arXiv preprint arXiv:1411.1784, 2014.
[32] P. Singh and N. Komodakis, "Cloud-gan: Cloud removal for sentinel-2 imagery using a cyclic consistent generative adversarial networks," in IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium, 2018: IEEE, pp. 1772-1775.
[33] J.-Y. Zhu, T. Park, P. Isola, and A. A. Efros, "Unpaired image-to-image translation using cycle-consistent adversarial networks," in Proceedings of the IEEE international conference on computer vision, 2017, pp. 2223-2232.
[34] C. Grohnfeldt, M. Schmitt, and X. Zhu, "A conditional generative adversarial network to fuse SAR and multispectral optical data for cloud removal from Sentinel-2 images," in IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium, 2018: IEEE, pp. 1726-1729.
[35] G. E. Hunt, "Radiative properties of terrestrial clouds at visible and infra‐red thermal window wavelengths," Quarterly Journal of the Royal Meteorological Society, vol. 99, no. 420, pp. 346-369, 1973.
[36] B. Huang, L. Zhi, C. Yang, F. Sun, and Y. Song, "Single satellite optical imagery dehazing using SAR image prior based on conditional generative adversarial networks," in Proceedings of the IEEE/CVF winter conference on applications of computer vision, 2020, pp. 1806-1813.
[37] A. Mehta, H. Sinha, M. Mandal, and P. Narang, "Domain-aware unsupervised hyperspectral reconstruction for aerial image dehazing," in Proceedings of the IEEE/CVF winter conference on applications of computer vision, 2021, pp. 413-422.
[38] A. Dosovitskiy et al., "An image is worth 16x16 words: Transformers for image recognition at scale," arXiv preprint arXiv:2010.11929, 2020.
[39] Z. Wang, X. Cun, J. Bao, W. Zhou, J. Liu, and H. Li, "Uformer: A general u-shaped transformer for image restoration," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 17683-17693.
[40] X. Qin, Z. Wang, Y. Bai, X. Xie, and H. Jia, "FFA-Net: Feature fusion attention network for single image dehazing," in Proceedings of the AAAI conference on artificial intelligence, 2020, vol. 34, no. 07, pp. 11908-11915.
[41] A. Mittal, A. K. Moorthy, and A. C. Bovik, "No-reference image quality assessment in the spatial domain," IEEE Transactions on image processing, vol. 21, no. 12, pp. 4695-4708, 2012.
[42] D. L. Ruderman, "The statistics of natural images," Network: computation in neural systems, vol. 5, no. 4, p. 517, 1994.
[43] N.-E. Lasmar, Y. Stitou, and Y. Berthoumieu, "Multiscale skewed heavy tailed model for texture analysis," in 2009 16th IEEE International Conference on Image Processing (ICIP), 2009: IEEE, pp. 2281-2284.
[44] B. Schölkopf, A. J. Smola, R. C. Williamson, and P. L. Bartlett, "New support vector algorithms," Neural computation, vol. 12, no. 5, pp. 1207-1245, 2000.
[45] A. Mittal, R. Soundararajan, and A. C. Bovik, "Making a “completely blind” image quality analyzer," IEEE Signal processing letters, vol. 20, no. 3, pp. 209-212, 2012.
[46] N. Venkatanath, D. Praneeth, M. C. Bh, S. S. Channappayya, and S. S. Medasani, "Blind image quality evaluation using perception based features," in 2015 twenty first national conference on communications (NCC), 2015: IEEE, pp. 1-6.
[47] F. Yang, H. Fan, P. Chu, E. Blasch, and H. Ling, "Clustered object detection in aerial images," in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 8311-8320.
[48] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, "Mobilenetv2: Inverted residuals and linear bottlenecks," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 4510-4520.
[49] D. Lin, G. Xu, X. Wang, Y. Wang, X. Sun, and K. Fu, "A remote sensing image dataset for cloud removal," arXiv preprint arXiv:1901.00600, 2019.
[50] O. Kupyn, V. Budzan, M. Mykhailych, D. Mishkin, and J. Matas, "Deblurgan: Blind motion deblurring using conditional adversarial networks," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 8183-8192.
[51] Y. Jiang et al., "Enlightengan: Deep light enhancement without paired supervision," IEEE transactions on image processing, vol. 30, pp. 2340-2349, 2021.
[52] R. Li, L.-F. Cheong, and R. T. Tan, "Heavy rain image restoration: Integrating physics model and conditional adversarial learning," in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 1633-1642.
[53] R. Li, J. Pan, Z. Li, and J. Tang, "Single image dehazing via conditional generative adversarial network," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 8202-8211.
[54] O. Ronneberger, P. Fischer, and T. Brox, "U-net: Convolutional networks for biomedical image segmentation," in Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18, 2015: Springer, pp. 234-241.
[55] P. Isola, J.-Y. Zhu, T. Zhou, and A. A. Efros, "Image-to-image translation with conditional adversarial networks," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1125-1134.
[56] A. Jolicoeur-Martineau, "The relativistic discriminator: a key element missing from standard GAN," arXiv preprint arXiv:1807.00734, 2018.
[57] J. Johnson, A. Alahi, and L. Fei-Fei, "Perceptual losses for real-time style transfer and super-resolution," in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part II 14, 2016: Springer, pp. 694-711.
[58] A. G. Howard et al., "Mobilenets: Efficient convolutional neural networks for mobile vision applications," arXiv preprint arXiv:1704.04861, 2017.
[59] L. Itti, C. Koch, and E. Niebur, "A model of saliency-based visual attention for rapid scene analysis," IEEE Transactions on pattern analysis and machine intelligence, vol. 20, no. 11, pp. 1254-1259, 1998.
[60] H. Larochelle and G. E. Hinton, "Learning to combine foveal glimpses with a third-order Boltzmann machine," Advances in neural information processing systems, vol. 23, 2010.
[61] Y. Zhang, K. Li, K. Li, L. Wang, B. Zhong, and Y. Fu, "Image super-resolution using very deep residual channel attention networks," in Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 286-301.
[62] L.-C. Chen, Y. Yang, J. Wang, W. Xu, and A. L. Yuille, "Attention to scale: Scale-aware semantic image segmentation," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 3640-3649.
[63] S. Woo, J. Park, J.-Y. Lee, and I. S. Kweon, "Cbam: Convolutional block attention module," in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 3-19.
[64] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770-778.
[65] T. Zhang, H.-M. Hu, and B. Li, "A naturalness preserved fast dehazing algorithm using HSV color space," IEEE Access, vol. 6, pp. 10644-10649, 2018.
[66] S. Bianco, L. Celona, F. Piccoli, and R. Schettini, "High-resolution single image dehazing using encoder-decoder architecture," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2019, pp. 0-0.
[67] I. Loshchilov and F. Hutter, "Decoupled weight decay regularization," arXiv preprint arXiv:1711.05101, 2017.
[68] L. N. Smith and N. Topin, "Super-convergence: Very fast training of neural networks using large learning rates," in Artificial intelligence and machine learning for multi-domain operations applications, 2019, vol. 11006: SPIE, pp. 369-386.
[69] B. Cai, X. Xu, K. Jia, C. Qing, and D. Tao, "Dehazenet: An end-to-end system for single image haze removal," IEEE transactions on image processing, vol. 25, no. 11, pp. 5187-5198, 2016.