HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES

HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES

Image matting often requires advanced image processing, especially in conditions, when small details such as hair are present in the image. In this article the hybrid method for human image matting based on convolutional neural network and principal curvatures is proposed. The U-Net based neural net...

Full description

Bibliographic Details
Main Authors: T. B. Sagindykov, E. A. Pavelyeva
Format: Article
Language:English
Published: Copernicus Publications 2021-04-01
Series:The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Online Access:https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIV-2-W1-2021/183/2021/isprs-archives-XLIV-2-W1-2021-183-2021.pdf
id doaj-9460fef6f3da40c9a40dacd11e700c2f
record_format Article
spelling doaj-9460fef6f3da40c9a40dacd11e700c2f2021-04-15T22:07:21ZengCopernicus PublicationsThe International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences1682-17502194-90342021-04-01XLIV-2-W1-202118318710.5194/isprs-archives-XLIV-2-W1-2021-183-2021HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATUREST. B. Sagindykov0E. A. Pavelyeva1Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, RussiaLaboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, RussiaImage matting often requires advanced image processing, especially in conditions, when small details such as hair are present in the image. In this article the hybrid method for human image matting based on convolutional neural network and principal curvatures is proposed. The U-Net based neural network is used to predict a rough foreground segmentation mask. Then the obtained foreground mask is refined by principal curvatures method to process the elongated hair-like structures. Test results show that the proposed method can improve the coarse human segmentation.https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIV-2-W1-2021/183/2021/isprs-archives-XLIV-2-W1-2021-183-2021.pdf
collection DOAJ
language English
format Article
sources DOAJ
author T. B. Sagindykov
E. A. Pavelyeva
spellingShingle T. B. Sagindykov
E. A. Pavelyeva
HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
author_facet T. B. Sagindykov
E. A. Pavelyeva
author_sort T. B. Sagindykov
title HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES
title_short HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES
title_full HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES
title_fullStr HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES
title_full_unstemmed HUMAN IMAGE MATTING BASED ON CONVOLUTIONAL NEURAL NETWORK AND PRINCIPAL CURVATURES
title_sort human image matting based on convolutional neural network and principal curvatures
publisher Copernicus Publications
series The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
issn 1682-1750
2194-9034
publishDate 2021-04-01
description Image matting often requires advanced image processing, especially in conditions, when small details such as hair are present in the image. In this article the hybrid method for human image matting based on convolutional neural network and principal curvatures is proposed. The U-Net based neural network is used to predict a rough foreground segmentation mask. Then the obtained foreground mask is refined by principal curvatures method to process the elongated hair-like structures. Test results show that the proposed method can improve the coarse human segmentation.
url https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIV-2-W1-2021/183/2021/isprs-archives-XLIV-2-W1-2021-183-2021.pdf
work_keys_str_mv AT tbsagindykov humanimagemattingbasedonconvolutionalneuralnetworkandprincipalcurvatures
AT eapavelyeva humanimagemattingbasedonconvolutionalneuralnetworkandprincipalcurvatures
_version_ 1721526051509633024

Similar Items