![]() Recently, denoising using a convolutional neural network (CNN) has been studied. Block-matching and 3-dimensional (3D) filtering (BM3D), referring to an enhanced version of NLM, show excellent performance, but the processing time can be very long due to high computational complexity. For denoising in single-view images, many studies based on the non-local mean (NLM) algorithm have been suggested. These research efforts have been extended from single-view image to videos and multi-view images. As a result, denoising is a prerequisite to produce multi-view-based image contents for next generation immersive media to satisfy both the expected image quality and the compression efficiency.Ī tremendous amount of works has been done in an effort to reduce noise in images. Noise decreases the temporal correlation of the pseudo sequence, consisting of multi-view images (hereinafter referred to as multi-view sequence) therefore, the compression efficiency is reduced and transmission in an insufficient network bandwidth becomes difficult. In addition, high-quality multi-view VR contents with a resolution exceeding 4 K require far more overhead for the storage and transmission of data compared to the conventional high-resolution images or the popular 360-degree video. When a user walks around in a light field or a multi-view based virtual space wearing a head-mounted display (HMD), the display is close to the eyes such that noise may severely deteriorate the quality of the experience. The above has been added to the manuscript. It takes a very long time to remove the noise generated in a large amount of multi-view. In particular, since the multi-view is photographed from multiple points or multiple cameras, the pattern of noise generated may be more diverse. This type of noise can arise in far more diverse patterns in multi-view images than in single-view images. Noise appears in various ways due to various causes, such as the amount of light reaching the camera sensor or reflecting off the dust in the air. During the image capturing step, noise can be generated for various reasons. ![]() Recently, light field-based virtual reality (VR) recognizes an interactive and immersive “walk-around space” by rendering an image of an arbitrary viewpoint through a combination of pixels. The light field information of dense multi-view images is frequently used in such applications as post-refocusing, super-resolution, occlusion reconstruction and depth estimation. ![]() Multi-view images or light field data, acquired from a multi-camera array by capturing a plurality of images of different viewpoints, are used in various applications, including the synthesis of a high-resolution image with a high dynamic range or the reconstruction of an occluded area through synthetic aperture photography (SAP). Experimental results show the proposed scheme significantly improves the compression efficiency of denoised views up to 76.05%, maintaining good denoising quality compared to the popular conventional denoise algorithms. Assuming the sequential processes of denoising and compression, multi-view geometry-based denoising is performed keeping the temporal correlation among views. In this paper, the structural characteristics of linear multi-view images are fully utilized to increase the denoising speed and performance as well as to improve the compression efficiency. Therefore, denoising is a prerequisite to produce multi-view-based image contents. However, noise can be easily generated during image capturing, and these noisy images severely deteriorate both the quality of experience and the compression efficiency. The compression efficiency is also critical because a large amount of multi-view data needs to be stored and transferred. High image quality is essential in systems with a near-eye display device. Immersive virtual reality is an important example. Multi-view or light field images have recently gained much attraction from academic and commercial fields to create breakthroughs that go beyond simple video-watching experiences.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |