使用殘餘變分自編碼器結合超解析深度網路之高品質瑕疵影像還原

High Quality Defective Image Reconstruction Using Residual VAE Combined with deep Super-Resolution Network

林光中(Guang-Chung Lin)

指導教授 ： 郭忠民

義守大學/智慧科技學院/資訊工程學系/碩士(2021年)

摘要

由於資訊科技的快速發展，多媒體內容尤其是影像的部分受到了廣泛的應用。然而來自特定來源的影像品質參差不齊，難以運用於不同的應用當中。例如紅外線影像的品質對於人類的視覺感受來說非常的差，因此想要對紅外線影像進行人臉辨識會非常的困難甚至難以達成。另外影像也可能會因人為加工處理而導致品質降低，例如二值化、馬賽克、模糊化等等，這些影像不只對機器是難以辨認，對於人類的視覺觀察也十分的困難，在本文中我們提出一種用於影像修復的殘餘自編碼器，這個方法可以很顯著的提升影像品質，並且成功提高辨識率。
測試結果證實了我們的預期，結果顯示了客觀品質以及主觀品質都有優異的表現。

關鍵字

辨識 ； 人類視覺 ； 深度殘餘變分自編碼器 ； 影像修復

並列摘要

Due to the rapid development of information technology, the multimedia content especially images are most widely used. However, the images that come from specific source are with various quality and hard to apply for different applications. For example, the quality of infrared images are very poor for human vision, therefore to perform infrared images for face recognition, it will be very difficult even impossible. On the other hands, images probably degrade by manmade such as binarization, mosaic, blurring. Those images are not only hard for human recognition, but also difficult for human vision and observation. In this work, we will propose a deep residual variational autoencoder for image recovery. The proposed method can improve the image visual quality significantly, and successfully increase the recognition rate.
Simulation results confirm our expectation, and show the excellent performance not only objective quality but subjective quality.

並列關鍵字

recognition ； human vision ； deep residual variational autoencoder ； image recovery

參考文獻

[1] G. Wang, J. C. Ye, K. Mueller and J. A. Fessler, "Image Reconstruction is a New Frontier of Machine Learning," in IEEE Transactions on Medical Imaging, vol. 37, no. 6, pp. 1289-1296, June 2018, doi: 10.1109/TMI.2018.2833635.

[2] S. Albawi, T. A. Mohammed and S. Al-Zawi, "Understanding of a convolutional neural network," 2017 International Conference on Engineering and Technology (ICET), 2017, pp. 1-6, doi: 10.1109/ICEngTechnol.2017.8308186.

[3] D. P. Kingma and M. Welling, Auto-encoding variational bayes, 2013.

[4] Akrami Haleh, Anand A Joshi, Jian Li and Richard M Leahy, Robust variational autoencoder, 2019.

[5] Pawlowski Nick, Matthew CH Lee, Martin Rajchl, Steven McDonagh, Enzo Ferrante, Konstantinos Kam-nitsas, Sam Cooke, Susan Stevenson, Aneesh Khetani, Tom Newman et al., "Unsupervised lesion detection in brain ct using bayesian convolutional autoencoders", OpenReview, 2018.