Yanzhu Zhang, Haishuai Zhang, Xiaomeng Zhang, J. Pu, Xiaoyan Wang
{"title":"基于轻量级对抗网络的光纤散斑恢复","authors":"Yanzhu Zhang, Haishuai Zhang, Xiaomeng Zhang, J. Pu, Xiaoyan Wang","doi":"10.1109/DDCLS52934.2021.9455515","DOIUrl":null,"url":null,"abstract":"When light with object information passes through a multi-core fiber, the speckle pattern is obtained. The reconstruction of the original image from the speckle pattern is crucial. In this paper, we propose a lightweight adversarial network for reconstruct image from the speckle pattern. Combining the characteristics of U-Net network and Mobile-Net, a lightweight Mobile-U-Net network is formed to reduce the number of network parameters by using deep separable convolution to realize fast reconstructing image. The adversarial network is also introduced to restrain the quality of the restored image and solve the quality problem of the restored image further. Thus, a high-quality reconstructing image can be achieved.","PeriodicalId":325897,"journal":{"name":"2021 IEEE 10th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"97 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Fiber Optic Speckle Recovery Based on Lightweight Adversarial Network\",\"authors\":\"Yanzhu Zhang, Haishuai Zhang, Xiaomeng Zhang, J. Pu, Xiaoyan Wang\",\"doi\":\"10.1109/DDCLS52934.2021.9455515\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"When light with object information passes through a multi-core fiber, the speckle pattern is obtained. The reconstruction of the original image from the speckle pattern is crucial. In this paper, we propose a lightweight adversarial network for reconstruct image from the speckle pattern. Combining the characteristics of U-Net network and Mobile-Net, a lightweight Mobile-U-Net network is formed to reduce the number of network parameters by using deep separable convolution to realize fast reconstructing image. The adversarial network is also introduced to restrain the quality of the restored image and solve the quality problem of the restored image further. Thus, a high-quality reconstructing image can be achieved.\",\"PeriodicalId\":325897,\"journal\":{\"name\":\"2021 IEEE 10th Data Driven Control and Learning Systems Conference (DDCLS)\",\"volume\":\"97 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2021-05-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2021 IEEE 10th Data Driven Control and Learning Systems Conference (DDCLS)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/DDCLS52934.2021.9455515\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 IEEE 10th Data Driven Control and Learning Systems Conference (DDCLS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/DDCLS52934.2021.9455515","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Fiber Optic Speckle Recovery Based on Lightweight Adversarial Network
When light with object information passes through a multi-core fiber, the speckle pattern is obtained. The reconstruction of the original image from the speckle pattern is crucial. In this paper, we propose a lightweight adversarial network for reconstruct image from the speckle pattern. Combining the characteristics of U-Net network and Mobile-Net, a lightweight Mobile-U-Net network is formed to reduce the number of network parameters by using deep separable convolution to realize fast reconstructing image. The adversarial network is also introduced to restrain the quality of the restored image and solve the quality problem of the restored image further. Thus, a high-quality reconstructing image can be achieved.