Free-viewpoint television (FTV) and virtual navigation appear to be hot research topics. In this paper, the authors study practical development of free-viewpoint systems that provide functionality horizontal around real scenes. The considerations are focused on use purely optical depth estimation might employed in next few years. architectures such discussed detail, including acquisition, preprocessing, estimation, compression, presentation. particular, optimization camera locations is...

This paper presents a new objective quality metric that was adapted to the complex characteristics of immersive video (IV) which is prone errors caused by processing and compression multiple input views virtual view synthesis. The proposed metric, IV-PSNR, contains two techniques allow for evaluation loss typical distortions: corresponding pixel shift global component difference. performed experiments compared proposal with 31 state-of-the-art metrics, showing their performance in assessment...

This paper presents the overview and rationale behind Decoder-Side Depth Estimation (DSDE) mode of MPEG Immersive Video (MIV) standard, using Geometry Absent profile, for efficient compression immersive multiview video. A MIV bitstream generated by an encoder operating in DSDE does not include depth maps. It only contains information required to reconstruct them client or cloud: decoded views metadata. The explains technical details techniques supported this novel mode. description...

The paper presents a new method of depth estimation, dedicated for free-viewpoint television (FTV) and virtual navigation (VN). In this method, multiple arbitrarily positioned input views are simultaneously used to produce maps characterized by high inter-view temporal consistencies. estimation is performed segments their size control the trade-off between quality processing time estimation. Additionally, an original technique proposed improvement consistency maps. This uses prediction...

In the paper, we propose a new method for virtual view synthesis called as Multiview Synthesis. our approach, views are synthesized using two neighboring real views, but disoccluded areas not inpainted, filled by information from further - additional traditionally included in synthesis. The whole is performed with triangles rather than individual pixels. proposed Synthesis, steps of adaptive color correction, blurred edge removal and spatial blurring also included. experimental results show...

This paper describes two methods for increasing the efficiency of MPEG Immersive Video (MIV) coding standard. The manipulate MIV-formatted atlas videos by considering principles widely used video encoders, enhancing encoding immersive content. first method, patch average color modification removes constant component all YCbCr components each within atlases, resulting in reduction number and magnitude edges a texture video. second proposed method changes dynamic range geometry (depth) atlas,...

We deal with the processing of multiview video acquired by use practical thus relatively simple acquisition systems that have a limited number cameras located around scene on independent tripods. The real-camera locations are nearly arbitrary as it would be required in real-world Free-Viewpoint Television systems. appropriate test sequences also reported. describe family original extensions and adaptations algorithms adapted to camera positions scene. techniques constitute chain for they...

In this article, we propose a depth map refinement method that increases the quality of immersive video. The proposal highly enhances inter-view consistency maps (estimated or acquired by any method), crucial for achieving required fidelity virtual view synthesis process. described method, only information from is used, as use texture can introduce errors in refinement, mostly due to color inconsistencies and noise. order evaluate performance compare it with state art, three experiments were...

The paper reports a new view synthesis method for omnidirectional video with the ability to slightly displace virtual viewpoint, i.e. describes novel 3DoF+ 360 video. This is noteworthy because of its high versatility and reliability: appropriate both perspective input views, able render produced synthetic views differ from respective ground truth images less than other methods. These important features result several innovations, e.g., prioritization efficient inpainting adapted...

The paper presents the concept of a practical free-viewpoint television system with purely optical depth estimation. consists camera modules that contain pairs or triples cameras together respective microphones. can be sparsely located in arbitrary positions around scene. Each module is equivalent to video sensor and hardware requirements, audio processing algorithms preliminary experimental results are reported. In particular, for such systems, compression technique discussed more efficient...

Accurate extrinsic parameters calibration is crucial particularly in immersive video, where camera plays a significant role, as its quality essential for accurate reconstruction and efficient compression of three-dimensional scenes. While methods intrinsic calibration, color correction, depth estimation are publicly available, there lack versatile techniques estimating extrinsics the context video. The proposed Extrinsic Camera Parameters Calibration (ECPC) software addresses these...

The paper considers the influence of a lossy compression multiview video on quality estimated depth maps. We have encoded set views with commonly known techniques used in compact IP cameras: HEVC (MPEG H part 2 / H.265), A VC 4 10 H.264) and MPEG H.262. Basing decoded data we maps rendered intermediate virtual views. drop was studied by comparison out from an uncompressed video. obtained results show that reasonable rate, lower is acceptable for real implementable FTV systems.

In this paper, the color correction method developed for immersive video systems is presented. The proposed significantly increases consistency of characteristics multiview sequences, understood both as temporal and inter-view consistency, what highly improves subjective quality synthesized virtual views presented to final user system. Moreover, proposal allows increase depth maps calculated natural e.g., with colors inconsistent due different lighting conditions. It enables more efficient...

The paper describes an experimental multiview video produc-tion, processing and delivery chain developed at Poznan Univer-sity of Technology for research on free-viewpoint television. multiview-video acquisition system consists HD camera units with wireless synchronization, control, stor-age power supply units. Therefore no cabling is needed in the system, which important shooting real-world events. mostly used nearly circular setup cameras but locations are arbitrary, procedures calibration...

Delays in the delivery of immersive video to Head Mounted Devices (HMDs) are considered paper. The goal this paper is analyze impact streaming latency on user experience with HMDs. reports results subjective quality assessment as a function delay. possible practical solutions identified and, base, an experimental model systems proposed. In order properly test influence delay quality, extensive tests have been performed. Interesting conclusions drawn, showing that human acceptance translation...

This paper presents a new depth estimation method for multiview systems with arbitrary camera locations. The exploits the graph cuts method, where vertices of represent segments used controlling trade-off between quality maps and time estimation, while preserving original resolution map. Moreover, inter-view consistency maps, crucial free-viewpoint television systems, is ensured by introduction suitable connections in optimized graph. It makes proposed first that allows generation...

In the article we deal with problem of camera positioning in sparse multiview systems applications to free navigation. The limited number cameras, though makes system relatively practical, implies problems proper depth estimation and virtual view synthesis, due increased amount occluded areas. We present experimental results for optimal depending on two factors - characteristics an acquired scene multi-camera (linear or circular setup). show correlation between occlusions a gain from using...

The demonstration is aimed at presenting a simple free-viewpoint television system that under development Poznań University of Technology, Poznań, Poland. Video acquisition using sparsely distributed pairs video cameras, the depth estimation from video, virtual rendering on server feature perspective FTV system. original and view synthesis algorithms ensure good quality views generated during walks around scene. developed for broadcasting sport (e.g. judo, karate, volleyball) cultural...

In this paper, a modification of the graph-based depth estimation is presented.The purpose proposed to increase quality estimated maps, reduce time estimation, and temporal consistency maps.The based on image segmentation using superpixels, therefore in first step previous frames used currently processed frame order overall estimation.In next step, map from optimization as initial values for current frame.It results better representation silhouettes objects maps reduced computational...

In free-viewpoint television, high-quality depth maps are substantial for a virtual view synthesis free navigation purposes. this paper, we propose new method of map quality improvement and resolution increase. Our is intended low acquired through estimation using multiview video. The proposed based on segmentation an high-resolution image. the estimated map, searching outliers performed in neighborhood created from similar segments view. Experimental results show high effectiveness...

In this paper, the authors describe two methods designed for reducing spatiotemporal redundancy of video within MPEG Immersive (MIV) encoder: patch occupation modification and cluster splitting. These allow optimizing important parameters immersive video: bitrate pixelrate. The method significantly decreases number active pixels texture depth produced by MIV encoder. Cluster splitting total area needed storing information from multiple input views, decreasing Both proposed paper were...

This paper presents the modified HEVC Screen Content Coding (SCC) that was adapted to be more efficient as an internal video coding of MPEG Immersive Video (MIV) codec. The basic, unmodified SCC is already known useful in such application. However, this paper, we propose three additional improvements increase efficiency immersive coding. First, analyze using quarter-pel accuracy intra block copy technique provide a effective search best candidate copied encoding process. second proposal use...