Improving colorization through denoiser with MLP
0
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
Colorizing black and white photos and movies enhances the old versions, and make them more interesting to see. Therefore, we became interested in image colorization, which plays an essential role in the field of computer vision. Recently, deep learning techniques for image colorization have progressed remarkably. In this study, colorization is carried out using a novel deep learning model for vivid coloring. The model consists of two components: Pix2Pix as a generative adversarial network (GAN), and the multi-layer perceptron (MLP) of a denoiser. The Pix2Pix primarily generates the color image for a given grayscale image as input, and the MLP transforms the colored image into a vivid color image by filtering out noise. In our experiments, the grayscale images used as input images were images of natural objects and artifacts. We observed that the predicted images provided as output images were more neatly colored through the proposed method.
Abstract
Colorizing black and white photos and movies enhances the old versions, and make them more interesting to see. Therefore, we became interested in image colorization, which plays an essential role in the field of computer vision. Recently, deep learning techniques for image colorization have progressed remarkably. In this study, colorization is carried out using a novel deep learning model for vivid coloring. The model consists of two components: Pix2Pix as a generative adversarial network (GAN), and the multi-layer perceptron (MLP) of a denoiser. The Pix2Pix primarily generates the color image for a given grayscale image as input, and the MLP transforms the colored image into a vivid color image by filtering out noise. In our experiments, the grayscale images used as input images were images of natural objects and artifacts. We observed that the predicted images provided as output images were more neatly colored through the proposed method.
참고문헌 (18)
- H. Bahng, "Coloring with words : Guiding image colorization through text-based palette generation", Proceedings of the European conference on Computer Vision (ECCV), pp.443~459 (2018)
- F. Baldassarre, "Deep koalarization: Image colorization using CNN and inceptionResNet-v2" (2017)
- S. Iizuka, "Let there be color!: Joint end-to-end learning of global and local image priors for automatic image colorization with simultaneous classification", ACM Transaction on Graphics (Proceedings of SIGGRAPH), 35(4), pp.1~11 (2016)
- G. Larsson, "Learning representations for automatic colorization", European Conference on Computer Vision 2016, pp.577~593 (2016)
- Z. Yu, "Retinal image synthesis from multiple-landmarks input with generative adversarial networks", Biomed Engineering Online, 18 (2019)
- I. Goodfellow, "Generative adversarial networks", 63(11), pp.139~144 (2020)
- I. J. Goodfellow, "Generative adversarial networks", Proceedings of the international conference on neural information processing systems (NIPS 2014), 3, pp.2672~2680 (2014)
- A. Radford, "Unsupervised representation learning with deep convolutional generative adversarial networks", ~ (2015)
- "Conditional generative adversarial nets"[URL]
- K. Armanious, "MedGAN:medical image translation using GANs", Computerized Medical Imaging Graphics, 79 (2020)
- P. Isola, "Image-toimage translation with conditional adversarial networks", pp.1125~1134 (2017)
- O. Ronneberger, "Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015", pp.234~241 (2015)
- Z. Wang, "Image quality assessment : From error visibility to structural similarity", IEEE Transaction on Image Processing, 13(4), pp.600~612 (2004)
- S. Treneska, "GAN-based image colorization for self-supervised visual feature learning", Sensors, 22(4), pp.1599~ (2022)[DOI]
- J. Zhang, "Computer Vision and Pattern Recognition" (2022)
- M. R. Joshi, "Auto-colorization of historical images using deep convolutional neural networks", Mathematics, 8(12), pp.2258~ (2020)[DOI]
- F. Baldassarre, "Deep koalarization: Image colorization using CNNs and inception-resnet-v2" (2017), KTH Royal Institute of Technology
- "Exploring convolutional neural networks for automatic image colorization" (2016)[URL]