Artigo completo - Open Access.

Idioma principal | Segundo idioma

IMAGE QUALITY ENHANCEMENT OF SCANNED PHOTOS:COMPARISON OF DEEP LEARNING TECHNIQUES

MELHORIA DA QUALIDADE DE IMAGEM DE FOTOS DIGITALIZADAS: COMPARAÇÃO DAS TÉCNICAS DE APRENDIZAGEM PROFUNDA

Santos, Victor Rocha ; Pagano, Tiago ; Kirstene , Lucas ; Ortega , Lucas ; Matos, Maíra ; Paranhos, José Vinícius ; Winkler, Ingrid ; Nascimento, Erick Giovani Sperandio ;

Artigo completo:

Currently, millions of photos are captured daily, and several factors caninfluence the quality of an image, causing distortions. Research has shown that thereare several ways to remove defects from images. This study aims to comparativelyanalyze the potential of Deep Learning techniques to improve scanned images withshadow, glare, crumpled paper, external lightning, change of perspective and wavedistortion defects. Based on a review of the literature on recent deep learningarchitectures, we have selected three, which were trained and refined to improve thequality of the images. The results indicate that the nets were able to attenuate andremove some defects. On this basis, these initial experiments demonstrate that deeplearning models are promising for the studied defects.

Artigo completo:

O objetivo deste trabalho é analisar comparativamente o potencial de técnicas de aprendizagem profunda para melhorar imagens digitalizadas com defeitos de sombra, reflexo de luz, papel amassado, luminosidade, mudança de perspectiva e ondulação. Com base em uma revisão da literatura sobre arquiteturas recentes de aprendizagem profunda, selecionamos três técnicas para remoção destes defeitos em imagens. Os resultados obtidos apontaram que as redes conseguiram atenuar alguns defeitos com intensidades variadas e, em alguns casos, removê-los. Conclui-se que estes experimentos iniciais demonstram que modelos de aprendizagem profunda são bastante promissores para a resolução de alguns dos defeitos estudados e que avanços significativos foram alcançados na melhoria da qualidade das imagens.

Palavras-chave: : Deep Learning; Image enhancement; Image denoising,

Palavras-chave: Aprendizagem profunda; Melhoria da qualidade de imagens;Remoção de ruídos em imagens,

DOI: 10.5151/siintec2020-IMAGEQUALITY

Referências bibliográficas
  • [1] O’MAHONY, N. et al. Deep learning vs. traditional computer vision. In: SPRINGER. Science and Information Conference. [S.l.], 2019. p. 128–144. 2 LI, X. et al. Blind geometric distortion correction on images through deep learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. [S.l.: s.n.], 2019. p. 4855–4864. 3 VU, T. et al. Fast and efficient image quality enhancement via desubpixel convolutional neural networks. In: Proceedings of the European Conference on Computer Vision (ECCV). [S.l.: s.n.], 2018. p. 0–0. VI INTERNATIONAL SYMPOSIUM ON INNOVATION AND TECHNOLOGY (SIINTEC) "Challenges in science, technology and innovation after COVID-19" ISSN: 2357-7592 4 CHIANG, J.-S. et al. Adaptive image enhancement method for document. In: IEEE.2017 International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS). [S.l.], 2017. p. 417–420. 5 BANNIGIDAD, P.; GUDADA, C. Restoration of degraded kannada handwritten paper inscriptions (hastaprati) using image enhancement techniques. In: IEEE.2017 International Conference on Computer Communication and Informatics (ICCCI). [S.l.],2017. p. 1–6. 6 RANI, N. S.; JAIN, A. S.; KIRAN, H. A unified preprocessing technique for enhancement of degraded document images. In: SPRINGER. International Conference on ISMAC in Computational Vision and Bio-Engineering. [S.l.], 2018. p. 221–233. 7 LIN, T.-Y. et al. Microsoft coco: Common objects in context. In: SPRINGER. European conference on computer vision. [S.l.], 2014. p. 740–755. 8 LECUN, Y. et al. Handwritten digit recognition with a back-propagation network. In: Advances in neural information processing systems. [S.l.: s.n.], 1990. p. 396–404. 9 LECUN, Y. et al. Object recognition with gradient-based learning. In: Shape,contour and grouping in computer vision. [S.l.]: Springer, 1999. p. 319–345. 10 TIAN, C.; XU, Y.; ZUO, W. Image denoising using deep cnn with batch renormalization. Neural Networks, Elsevier, v. 121, p. 461–473, 2020. 11 HE, K. et al. Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition. [S.l.: s.n.], 2016. p.770–778. 12 WANG, J.; HE, H.; PROKHOROV, D. V. A folded neural network autoencoder for dimensionality reduction. Procedia Computer Science, Elsevier, v. 13, p. 120–127, 2012. 13 GONDARA, L. Medical image denoising using convolutional denoising autoencoders. In: IEEE.2016 IEEE 16th International Conference on Data Mining Workshops (ICDMW). [S.l.], 2016. p. 241–246. 14 TJANDRA, A. et al. Stochastic gradient variational bayes for deep learning-based asr. In: IEEE.2015 IEEE Workshop on Automatic Speech Recognition and Understanding(ASRU). [S.l.], 2015. p. 175–180. 15 XU, Q. et al. The difference learning of hidden layer between autoencoder and variational autoencoder. In: IEEE.2017 29th Chinese Control And Decision Conference(CCDC). [S.l.], 2017. p. 4801–4804.
Como citar:

Santos, Victor Rocha ; Pagano, Tiago ; Kirstene , Lucas ; Ortega , Lucas ; Matos, Maíra ; Paranhos, José Vinícius ; Winkler, Ingrid ; Nascimento, Erick Giovani Sperandio ; "IMAGE QUALITY ENHANCEMENT OF SCANNED PHOTOS:COMPARISON OF DEEP LEARNING TECHNIQUES", p. 626-634 . In: Anais do VI Simpósio Internacional de Inovação e Tecnologia. São Paulo: Blucher, 2020.
ISSN 2357-7592, ISBN: 2357-7592
DOI 10.5151/siintec2020-IMAGEQUALITY

últimos 30 dias | último ano | desde a publicação


downloads


visualizações


indexações