{"title":"基于改进SCT的室内场景微光图像增强模型","authors":"Yuxiang Cheng, Chong Yang","doi":"10.1109/TOCS56154.2022.10016073","DOIUrl":null,"url":null,"abstract":"For the low-light enhancement model SCT, lowlight enhancement in an indoor environment, there will be some textures that are usually not characteristics of the object. These fake textures can generate misunderstandings. An improved SCT model is proposed in response to this problem, called SCT++. The improved model adds the spatial uniform loss function, the color constant loss function, and the brightness smoothing loss function. Comparing the SCT model and the SCT++ model through the LOL dataset, it is found that the convergence rate of the SCT++ model is greater than that of the SCT model, and the convergence rate is improved. Finally, SCT++, SCT, DCE++, GAN, LIME, LLVE, and RUAS were compared, and PSNR, SSIM, NIQE, LPIPS, MSE, and other image evaluation methods were selected. It is concluded that the SCT++ model is better than the SCT model in terms of PSNR, LPIPS, and MSE.","PeriodicalId":227449,"journal":{"name":"2022 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Low-light image enhancement model based on improved SCT in indoor scenes\",\"authors\":\"Yuxiang Cheng, Chong Yang\",\"doi\":\"10.1109/TOCS56154.2022.10016073\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"For the low-light enhancement model SCT, lowlight enhancement in an indoor environment, there will be some textures that are usually not characteristics of the object. These fake textures can generate misunderstandings. An improved SCT model is proposed in response to this problem, called SCT++. The improved model adds the spatial uniform loss function, the color constant loss function, and the brightness smoothing loss function. Comparing the SCT model and the SCT++ model through the LOL dataset, it is found that the convergence rate of the SCT++ model is greater than that of the SCT model, and the convergence rate is improved. Finally, SCT++, SCT, DCE++, GAN, LIME, LLVE, and RUAS were compared, and PSNR, SSIM, NIQE, LPIPS, MSE, and other image evaluation methods were selected. It is concluded that the SCT++ model is better than the SCT model in terms of PSNR, LPIPS, and MSE.\",\"PeriodicalId\":227449,\"journal\":{\"name\":\"2022 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS)\",\"volume\":\"41 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-12-11\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2022 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/TOCS56154.2022.10016073\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/TOCS56154.2022.10016073","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Low-light image enhancement model based on improved SCT in indoor scenes
For the low-light enhancement model SCT, lowlight enhancement in an indoor environment, there will be some textures that are usually not characteristics of the object. These fake textures can generate misunderstandings. An improved SCT model is proposed in response to this problem, called SCT++. The improved model adds the spatial uniform loss function, the color constant loss function, and the brightness smoothing loss function. Comparing the SCT model and the SCT++ model through the LOL dataset, it is found that the convergence rate of the SCT++ model is greater than that of the SCT model, and the convergence rate is improved. Finally, SCT++, SCT, DCE++, GAN, LIME, LLVE, and RUAS were compared, and PSNR, SSIM, NIQE, LPIPS, MSE, and other image evaluation methods were selected. It is concluded that the SCT++ model is better than the SCT model in terms of PSNR, LPIPS, and MSE.