scholarly journals 3D video conversion system based on depth information extraction

2018 ◽  
Vol 232 ◽  
pp. 02048
Author(s):  
Yingchun Zhang ◽  
Jianbo Huang ◽  
Siwen Duan
Keyword(s):  
Information Extraction ◽  
3D Video ◽  
Video Frame ◽  
Depth Information ◽  
Conversion System ◽  
Information Generation ◽  
3D Movie ◽  
Image Depth

3D movies have received more and more attention in recent years. However, the investment in making 3D movies is high and difficult, which restricts its development. And there are many existing 2D movie resources, and how to convert it into 3D movies is also a problem. Therefore, this paper proposes a 3D video conversion system based on depth information extraction. The system consists of four parts: segmentation of movie video frame sequences, extraction of frame image depth information, generation of virtual multi-viewpoint and synthesis of 3D video. The system can effectively extract the depth information of the movie and by it finally convert a 2D movie into a 3D movie.

A New Approach of 2D-to-3D Video Conversion and Its Implementation on Embedded System

2011 ◽  
Vol 58-60 ◽  
pp. 2552-2557
Author(s):  
Hao Jun Li ◽  
Yue Sheng Zhu
Keyword(s):  
Embedded System ◽  
Prediction Method ◽  
3D Video ◽  
Video Sequences ◽  
Video Frame ◽  
Depth Information ◽  
Video Scene ◽  
Camera Position ◽  
The Relationship ◽  
2D To 3D

Rather than based on the stereo vision principle and the relationship between the camera position and the video scene objects used in most of current 2D-to-3D video conversion algorithms, a new prediction method of video depth information based on video frame differences is proposed and implemented on an embedded platform in this paper. 3D stereoscopic video sequences are generated by using the original 2D video sequences and the depth information. The theoretical analysis and experimental results have showed that the proposed method is more feasible and efficiency compared with the current algorithms.

scholarly journals DEEP LEARNING FOR MONOCULAR DEPTH ESTIMATION FROM UAV IMAGES

2020 ◽  
Vol V-2-2020 ◽  
pp. 451-458
Author(s):  
L. Madhuanand ◽  
F. Nex ◽  
M. Y. Yang
Keyword(s):  
Deep Learning ◽  
Ground Level ◽  
Depth Estimation ◽  
Aerial Images ◽  
Aerial Image ◽  
Depth Information ◽  
Single Image ◽  
Monocular Depth ◽  
Uav Images ◽  
Image Depth

Abstract. Depth is an essential component for various scene understanding tasks and for reconstructing the 3D geometry of the scene. Estimating depth from stereo images requires multiple views of the same scene to be captured which is often not possible when exploring new environments with a UAV. To overcome this monocular depth estimation has been a topic of interest with the recent advancements in computer vision and deep learning techniques. This research has been widely focused on indoor scenes or outdoor scenes captured at ground level. Single image depth estimation from aerial images has been limited due to additional complexities arising from increased camera distance, wider area coverage with lots of occlusions. A new aerial image dataset is prepared specifically for this purpose combining Unmanned Aerial Vehicles (UAV) images covering different regions, features and point of views. The single image depth estimation is based on image reconstruction techniques which uses stereo images for learning to estimate depth from single images. Among the various available models for ground-level single image depth estimation, two models, 1) a Convolutional Neural Network (CNN) and 2) a Generative Adversarial model (GAN) are used to learn depth from aerial images from UAVs. These models generate pixel-wise disparity images which could be converted into depth information. The generated disparity maps from these models are evaluated for its internal quality using various error metrics. The results show higher disparity ranges with smoother images generated by CNN model and sharper images with lesser disparity range generated by GAN model. The produced disparity images are converted to depth information and compared with point clouds obtained using Pix4D. It is found that the CNN model performs better than GAN and produces depth similar to that of Pix4D. This comparison helps in streamlining the efforts to produce depth from a single aerial image.

scholarly journals Semantic Information Generation from Classification and Information Extraction

2004 ◽  
pp. 573-574
Author(s):  
Tércio de Morais Sampaio Silva ◽  
Frederico Luiz Gonçalves de Freitas ◽  
Rafael Cobra Teske ◽  
Guilherme Bittencourt
Keyword(s):  
Information Extraction ◽  
Semantic Information ◽  
Information Generation

A real-time 1080p 2D-to-3D video conversion system

2011 ◽  
Vol 57 (2) ◽  
pp. 915-922 ◽  
Author(s):  
Sung-Fang Tsai ◽  
Chao-Chung Cheng ◽  
Chung-Te Li ◽  
Liang-Gee Chen
Keyword(s):  
Real Time ◽  
3D Video ◽  
Conversion System ◽  
2D To 3D

Hardware/Software Co-Design of 2D-to-3D Video Conversion on FPGA

2013 ◽  
Vol 284-287 ◽  
pp. 3230-3234
Author(s):  
Thomas Schumann ◽  
Herbert Krauß ◽  
Yeong Kang Lai ◽  
Yu Fan Lai
Keyword(s):  
Image Inpainting ◽  
3D Video ◽  
View Synthesis ◽  
Image Warping ◽  
Depth Information ◽  
3D Image ◽  
Hole Filling ◽  
Synthesis Algorithm ◽  
3D Image Warping ◽  
2D To 3D

With advances in technology, 3D video technology becomes possible and attractive. However, there are still many pre-recorded 2D videos/images which need to get transferred to 3D. Hence this paper presents a high quality view synthesis algorithm and architecture for 2D-to-3D video conversion. During the process of view synthesis, the monocular depth information together with the intermediate view is synthesized to the left-eye and right-eye view. The proposed view synthesis algorithm consists of two parts: 3D image warping and inpainting (hole filling). 3D image warping transforms a 2D camera image plane to a 3D coordinate plane. However the integer grid points of the reference are warped to irregularly spaced points in the virtual view, resulting in occlusion problems. Thus inpainting is needed to fix the virtual images. The proposed algorithm shows an improved PSNR gain of 0.2~1.5dB. We adopt hardware/software co-design to accomplish the proposed view synthesis algorithm. For this we implemented the image inpainting on a FPGA device and the remaining algorithm in software.

A real-time time-consistent 2D-to-3D video conversion system using color histogram

2015 ◽  
Vol 61 (4) ◽  
pp. 524-530 ◽  
Author(s):  
Zhen Zhang ◽  
Shouyi Yin ◽  
Leibo Liu ◽  
Shaojun Wei
Keyword(s):  
Real Time ◽  
3D Video ◽  
Color Histogram ◽  
Conversion System ◽  
2D To 3D
Sign in / Sign up

Export Citation Format

Share Document