Skip to content
Free planNo credit card requiredStart writing
AI for Research home
AI for Research home
Dashboard
Sign inGet started

Search 250 million research papers

Read the abstract, see who wrote it and who cites it, and get the citation in APA, MLA, Chicago or Harvard.

About 3,037 papers for “Video post-processing”

Video post processing architectures

A. Beric · TU/e Research Portal · 2008

This PhD thesis is not the work of a single man.Many people contributed in many ways.

Cited by 4Open access

Video Post-processing Using DNNs

Power, Garbhan · Trinity's Access to Research Output (TARA) (Trinity College Dublin) · 2021

With an ever-increasing demand for online video streaming, faster and more efficient methods must constantly be developed to deliver high-quality content to consumers. With the current advances in Neural Networking, the use of a Deep Neural Network for media encoding is gaining significant interest from researchers. This report will analyse the possibility of using an auto-encoder Neural Network design for video post-processing. Expanding on the architecture described in Exploring DNNs for Media Compression two models are proposed. These models use a series of skip connections, which carry information forward through the network, as well as 3D convolutional layers, which use multiple frames to provide extra information to the network. The first proposed architecture is the MC model, which applies Motion Compensation (MC) to the previous frame in order to ensure that the information used during the 3D convolution is relevant to the pixels at the same location in the current frame. However, due to the additional computational power required to perform this motion compensation, the model was limited in terms of testing. It was tested for 24 epochs with a dataset of H.264 encoded media containing 2342 frames. The second model is the non-MC model, which does not apply motion compensation, is intended to better capture the motion using larger convolutions. This model was tested over 72 epochs with a dataset of 4622 H.264 encoded frames. In both cases, the media was encoded at multiple CRF values and subjected to data augmentation. Performance was assessed using MSE, PSNR and MS-SSIM, as well as the visual appearance. Both models showed disappointing performance, with media showing a drop in quality when post-processed. This was due to the media being subjected to a heavy amount of distortion in the chrominance and luminance. This was caused by the loss function, which minimised the MSE without ensuring that the YUV channels remained somewhat consistent. However, the motion compensation algorithm showed promising results, as it modelled the inter-frame movement better than both the non-MC model and the model proposed by R Trew. Less shadowing was present due to motion and the SSIM of post-processed media with high movement showed a significant drop. By implementing a more appropriate loss function, as well as more sophisticated form of motion estimation that uses pixel interpolation, there is confidence that a high-performing post-processor can be created. Furthermore, this project has highlighted various other avenues of exploration, such as using a network structure that has seen success for other authors, such as the multiframe models used in FlowNet, or by expanding the experiments to more recent codecs, such as AV1.

Cited by 0Open access

Preparing for post-catastrophe video processing

Anton van den Hengel, Henry Detmold, Anthony Dick et al. · 2007

ABSTRACT: The spread of video surveillance systems and video phones means that video cameras are more ubiquitous than ever before. In the event of a catastrophe the video captured by these cameras is an important source of information for those directing the response. This information is currently either ignored because it is seen as inaccessible, filtered through the media, or requires a huge commitment of human resources in order to perform the required processing. We present here an approach towards acquiring and processing this video in order to extract as much value as possible within the shortest time period. The approach presented allows flexible, intelligent video processing to be carried out quickly and securely. BIOGRAPHY: The authors are at the University of Adelaide, Australia. Dr. van den Hengel is director of the Australian Centre for Visual Technologies, and conducts research in automated video surveillance in collaboration with authors Dick and Hill. Dr. Detmold has extensive experience in software architecture research, and is currently working on scalable software architectures for large scale video surveillance networks. 1

Cited by 0

Efficient Implementation Of Video Post-Processing

Algorithms On The, Doina Petrescu · 2007

Deblocking and deringing are two video post-processing techniques largely used to remove coding artifacts and improve the visual quality when rendering low bit rate coded video. The algorithms used to achieve these tasks are computationally intensive and usually require high speed processors to be able to run in real time. Efficient implementations of signal adaptive filters for video post-processing can be obtained using the specialized features of the parallel BOPS DSP cores. The performance achieved by deblocking and deringing CIF and SDTV size video sequences on the MANTA^TM prototype chip are illustrated. It is shown that such complex tasks may be executed at low clock rates using the BOPS ManArray^TM technology.

Cited by 0

A video pre/post-processing LSI for video capture

T. Kinugasa, A. Nishizawa, K. Koshio et al. · IEEE Transactions on Consumer Electronics · 1996

A video processing LSI which converts video signals to digital YUV signals in various image formats was successfully developed for a video capture system. This LSI is suitable for PC cameras, camera/recorders and PC video capture boards.

Cited by 5

Image/video processing techniques : header compression and post-processing

Rare & Special e-Zone (The Hong Kong University of Science and Technology) · 2010

Image/video processing techniques : header compression and post-processing by Wen Yang thesis 2010 xii, 89 p. : ill. ; 30 cm Multimedia which has been in existence for centuries has greatly impacted the world. The popularity of multimedia in our…Read more ›

Cited by 0

Post-processing of compressed video using a unified metric for digital video processing

Lilla Boroczky, Yibin Yang · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2004

In this paper we propose a novel, post-processing system for compressed video sources. The proposed system explores the interaction between artifact reduction and sharpness/resolution enhancement to achieve optimal video quality for compressed (e.g. MPEG-2) sources. It is based on the Unified Metric for Digital Video Processing (UMDVP), which adaptively controls the post-processing algorithms according to the coding characteristics of the decoded video. The experiments carried out on several MPEG-2 encoded video sequences have shown significant improvement in picture quality compared to a system without the UMDVP control and to a system that did not exploit the interaction between artifact reduction and video enhancement. The UMDVP as well the proposed post-processing system can be easily adapted for different coding standard, such as MPEG-4, H.26x.

Cited by 0

Video Coding with Super-Resolution Post-Processing

Satoshi Kondo, Tadamasa Toma · 2006

We propose a video coding technique using super-resolution image processing. The proposed method reduces spatial resolution of specific pictures in input video sequence at first, namely it produces a mixed-resolution video sequence. And then the proposed method encodes the mixed-resolution video sequence. The spatially reduced pictures are converted to original resolution pictures with a super-resolution image processing. The proposed method can reduce the bit rate compared to conventional methods such as H.264, since it encodes the video sequence in which several pictures are spatially reduced. Moreover, the proposed method can prevent deterioration in subjective quality, since the spatially reduced pictures are restored with the super-resolution image processing. We evaluate the proposed method by implementing it on an H.264-based video coding scheme. Experimental results confirm that compared to H.264, our implementation of the proposed method can achieve a 3.5 dB improvement in PSNR at a low bit rate.

Cited by 6

New editing techniques for video post-processing

Volker Scholz · Publications of the UdS (Saarland University) · 2007

This thesis contributes to capturing 3D cloth shape, editing cloth texture and altering object shape and motion in multi-camera and monocular video recordings. We propose a technique to capture cloth shape from a 3D scene flow by determining optical flow in several camera views. Together with a silhouette matching constraint we can track and reconstruct cloth surfaces in long video sequences. In the area of garment motion capture, we present a system to reconstruct time-coherent triangle meshes from multi-view video recordings. Texture mapping of the acquired triangle meshes is used to replace the recorded texture with new cloth patterns. We extend this work to the more challenging single camera view case. Extracting texture deformation and shading effects simultaneously enables us to achieve texture replacement effects for garments in monocular video recordings. Finally, we propose a system for the keyframe editing of video objects. A color-based segmentation algorithm together with automatic video inpainting for filling in missing background texture allows us to edit the shape and motion of 2D video objects. We present examples for altering object trajectories, applying non-rigid deformation and simulating camera motion.

Cited by 0Open access

Efficient post-processing for block-based compressed video

Yongku Kim, T. Yi · 2004

In digital video coding, the 2-D block-based discrete cosine transform (DCT) has been widely used, combined with the quantization techniques of DCT coefficients, for the purpose of compression of video data. Video compression algorithms using this block-based transform coding method cause the blocking and ringing artifacts in the reconstructed images of a video sequence. In this paper, we propose the efficient post-processing algorithm for coded video sequence, where the ringing effects near sharp edges are smoothed out and the blocking effects are removed without blurring the edge regions at the decoder. The simulation examples of processing images coded by MPEG-2 are presented to demonstrate that a significant improvement has been achieved.

Cited by 2

Interactive rotoscope system "IMANA" for video post-processing

Kruse, S. · Publikationsdatenbank der Fraunhofer-Gesellschaft (Fraunhofer-Gesellschaft) · 1999

Describes the use of segmentation masks derived from conventional video sequences to create special effects for film, TV and video production. This system is termed IMANA (Image Analyser) and is based on automatically segmented layer masks. The author presents segmentally modified images and refers to problems of movement estimation.

Cited by 0

Video Coding Scheme Using Pre-Processing Degradation and Post-Processing Restoration

Han Wang, Luheng Jia, Haoqiang Ren et al. · 2023

Extending coding tools beyond but compatible to video coding standards provides a new path to higher coding efficiency. In this paper, we propose an efficient video coding scheme with pre-processing degradation and post-processing restoration. Proper amount of high frequency information is removed before encoding using edge-preserving filter, while the post-processing module exploiting convolutional neural network (CNN) restores degraded frames due to pre-processing and coding artifacts such as quantization error. Experimental results demonstrate that the proposed video coding scheme achieves coding efficiency gain of up to - 9.6% Bjontegaard delta bit rate (BDBR) compared with H.265/HEVC reference model.

Cited by 2

Versatile Video Coding-Post Processing Feature Fusion: A Post-Processing Convolutional Neural Network with Progressive Feature Fusion for Efficient Video Enhancement

T. S. Das, Xilong Liang, Kiho Choi · Applied Sciences · 2024

Advanced video codecs such as High Efficiency Video Coding/H.265 (HEVC) and Versatile Video Coding/H.266 (VVC) are vital for streaming high-quality online video content, as they compress and transmit data efficiently. However, these codecs can occasionally degrade video quality by adding undesirable artifacts such as blockiness, blurriness, and ringing, which can detract from the viewer’s experience. To ensure a seamless and engaging video experience, it is essential to remove these artifacts, which improves viewer comfort and engagement. In this paper, we propose a deep feature fusion based convolutional neural network (CNN) architecture (VVC-PPFF) for post-processing approach to further enhance the performance of VVC. The proposed network, VVC-PPFF, harnesses the power of CNNs to enhance decoded frames, significantly improving the coding efficiency of the state-of-the-art VVC video coding standard. By combining deep features from early and later convolution layers, the network learns to extract both low-level and high-level features, resulting in more generalized outputs that adapt to different quantization parameter (QP) values. The proposed VVC-PPFF network achieves outstanding performance, with Bjøntegaard Delta Rate (BD-Rate) improvements of 5.81% and 6.98% for luma components in random access (RA) and low-delay (LD) configurations, respectively, while also boosting peak signal-to-noise ratio (PSNR).

Cited by 5Open access

Coupled pre-/post-processing filters for predictive video coding

Kiran Misra, Joel Solé, Xiaoan Lu et al. · 2010

Pre-/post-processing filters obtained from lapped transforms achieve higher coding efficiency by exploiting transform block boundary correlation. Existing approaches however do not explicitly consider the impact of pre-filtering on prediction efficiency, thereby resulting in a lower overall rate-distortion performance. We present a framework where filters are designed with an objective function based on a prediction model. The post-filter is constructed so as to reverse the pre-filter processing at the decoder. We show how the pre-filters thus derived can be refined and local adaptation parameters obtained for better compression efficiency. To further enhance the performance, the mode decision is altered by separating the contribution of luma and chroma components in the rate-distortion model. The overall system performance is measured using a weighted Bjontegaard delta bitrate, showing an overall saving of 2.58% to 6.78% with respect to the “Key Technical Area” (KTA) codec. Additionally, the designed filters significantly reduce the blocking artifacts observed in KTA.

Cited by 0