Topic
Quantization (image processing)
About: Quantization (image processing) is a research topic. Over the lifetime, 7977 publications have been published within this topic receiving 126632 citations.
Papers published on a yearly basis
Papers
More filters
07 Dec 1998
TL;DR: This approach simply re-applies JPEG to the shifted versions of the already-compressed image, and forms an average, which offers better performance than other known methods, including those based on nonlinear filtering, POCS, and redundant wavelets.
Abstract: A novel method is proposed for post-processing of JPEG-encoded images, in order to reduce coding artifacts and enhance visual quality. Our method simply re-applies JPEG to the shifted versions of the already-compressed image, and forms an average. This approach, despite its simplicity, offers better performance than other known methods, including those based on nonlinear filtering, POCS, and redundant wavelets.
51 citations
Patent•
25 Apr 2005TL;DR: In this paper, an image information conversion apparatus and method by which picture quality deterioration caused by conversion from inputted MPEG2 image compression information into MPEG4 image compression to be outputted is prevented.
Abstract: The invention provides an image information conversion apparatus and method by which picture quality deterioration caused by conversion from inputted MPEG2 image compression information into MPEG4 image compression information to be outputted is prevented. When an I picture of an MPEG2 bit stream is to be converted into a P-VOP of an MPEG4 bit stream based on an estimated value of the complexity for each VOP, a scene change detection section detects whether or not the I picture includes a scene change. If a scene change is detected by the scene change detection section, then a GOV structure determination section determines that conversion of the I picture of the MPEG2 bit stream into a P-VOP of an MPEG4 bit stream should not be performed.
51 citations
TL;DR: The RD-OPT algorithm for DCT quantization optimization, which can be used as an efficient tool for near-optimal rate control in DCT-based compression techniques, such as JPEG and MPEG, is described.
Abstract: We describe the RD-OPT algorithm for DCT quantization optimization, which can be used as an efficient tool for near-optimal rate control in DCT-based compression techniques, such as JPEG and MPEG. RD-OPT measures DCT coefficient statistics for the given image data to construct rate/distortion-specific quantization tables with nearly optimal tradeoffs.
51 citations
01 Jun 2021
TL;DR: XLSR as mentioned in this paper is a hardware limitation aware, extremely lightweight quantization robust real-time super resolution network (xLSR) based on root modules introduced in Image classification.
Abstract: Single-Image Super Resolution (SISR) is a classical computer vision problem and it has been studied for over decades. With the recent success of deep learning methods, recent work on SISR focuses solutions with deep learning methodologies and achieves state-of-the-art results. However most of the state-of-the-art SISR methods contain millions of parameters and layers, which limits their practical applications. In this paper, we propose a hardware (Synaptics Dolphin NPU) limitation aware, extremely lightweight quantization robust real-time super resolution network (XLSR). The proposed model’s building block is inspired from root modules introduced in [15] for Image classification. We successfully applied root modules to SISR problem, further more to make the model uint8 quantization robust we used Clipped ReLU at the last layer of the network and achieved great balance between reconstruction quality and runtime. Furthermore, although the proposed network contains 30x fewer parameters than VDSR [16] its performance surpasses it on Div2K validation set. The network proved itself by winning Mobile AI 2021 Real-Time Single Image Super Resolution Challenge.
51 citations
20 Aug 2006
TL;DR: It is observed that the contourlet based algorithm outperforms wavelet and DCT based methods in this application and gives a distinct advantage over conventional techniques in images like maps which consist of lot of curves and texts.
Abstract: Digital watermarking has been proposed as a method of copyright protection of audio, images, video and text. We propose to use the newly introduced transform for two dimensional signals, namely, the contourlet transform for image watermarking application. We have carried out simulations with two kinds of images - maps consisting of lot of curves and texts and those with textures. The watermarked images were subjected to different attacks like mean filtering, quantization and JPEG compression. It is observed that the contourlet based algorithm outperforms wavelet and DCT based methods in this application. It is also apparent from the simulation results that contourlet based techniques give a distinct advantage over conventional techniques in images like maps which consist of lot of curves and texts
51 citations