Search

Deblocking filtering (DF) plays a major role in improving the visual quality for image and video coding. A well-designed DF scheme should achieve a good tradeoff between coding efficiency and coding complexity. The DF algorithm for multiview video coding (MVC) has adopted H.264 DF scheme to improve the video quality, which, however, leads to relatively high coding complexity. In this paper, we propose a low complexity DF algorithm in consideration of view correlation, which exploits the view correlation to skip the boundary strength (BS) computation of macroblocks (MBs) that are coded in motion skip mode. Experimental results show that the proposed algorithm can dramatically reduce DF time (35%-6%) without degrading the visual quality in comparison to the conventional algorithm.

Blocking artifacts are major obstacles in block-based transform coding (BTC) for low bit-rate mobile video. In this paper, we present a locally adaptive deblocking technique based on block activities around block boundaries in order to improve the visual quality of mobile video. Preprocessing makes an appropriate filtering mode chosen by removing undesired signals. The proposed filter consists of four modes depending on the types of the artifacts. Experimental results show the effectiveness of the proposed scheme compared with conventional deblocking filters including the H.264/AVC post-filter with regard to objective and subjective performance. The proposed filter improves PSNR up to 0.8dB compared with the non post-processed video and outperforms conventional filters with reduced complexity compared with the MPEG-4 deblocking filter.

In the perceptual coding scheme, the properties of the human visual system (HVS) are usually used to improve the coding efficiency. The frequency sensitivity is one of the most important properties of HVS and is usually used to improve the subjective quality. The frequency weighting technique is adopted in many video coding standards, such as MP EG 2 and H.264/AVC High Profile. However, the frequency weighting algorithms used in those standards are all in picture level. In this paper, a novel macroblock (MB)- level adaptive frequency weighting (MBAFW) algorithm is investigated. Compared with the picture-level adaptive frequency weighting (PAFW) algorithms, the encoder with MBAFW can select different frequency weighting strategies and different quantization matrices for each MB. The spatial context and the motion activities of one picture are both taken into consideration, which makes the frequency weighting adaptation more efficient and reasonable. The experimental results show that MBAFW can improve the subjective quality significantly. Additionally, MBAFW algorithm can achieve 10%~15% bitrate reductions with almost the same subjective quality.

Binaural cue coding (BCC) was introduced as an efficient representation method for MPEG-4 SAC (spatial audio coding). However, in a low bit-rate environment, the spectrum of BCC output signals degrades with respect to the perceptual level. The proposed system in this paper estimates VSLI (virtual source location information) as the side information. The VSLI is the angle representation of spatial images between channels on playback layout. The subjective assessment results show that the proposed method provides better audio quality than the BCC method for encoding multi-channel signals.

This paper presents a perceptual two-pass VBR MPEG-2 video encoder suitable for digital storage applications. The proposed algorithm aims to achieve perceptually uniform distortion among adjacent macroblocks within a frame and smoothen the inter-frame visual quality fluctuation as soon as possible. In the first-pass coding, we analyze the video sequence and collect some useful coding parameters to accurately measure the frame coding complexity. In the second-pass coding, the target bits are allocated to all frames optimally according to the coding complexity, certainly with decoder buffer considered. The human visual system characteristics including both the spatial and temporal masking effects are employed to improve the proposed adaptive quantization algorithm. Furthermore, an improved iteration based Qp selection algorithm is proposed to search for the optimal frame-level reference Qp resulting in satisfactory iteration speed and iteration precision. Experimental results show that the proposed algorithm can achieve superior visual quality both objectively and subjectively compared with the reference algorithms.

This paper examines the efficiency of MPEG-2 coding for interlaced and progressive video, and compares de-interlacing and picture rate up-conversion before and after coding. We found receiver side de-interlacing and picture rate up-conversion (i.e. after coding) to give better image quality at a given data rate. In contrast with some other publications, we found interlaced video coding to be better than progressive video coding for many relevant sequences, even when comparing the results on a progressive display.

We present a novel error concealment method for MPEG-2 video decoders. Imperfect transmission of block-based compressed images may result in loss of blocks, making image degradation inevitable. In the hybrid error concealment method, both spatial and temporal error concealment repair the damaged regions through adaptive interpolation in the respective domains. In order for the hybrid method to yield good performance it should be presumed that coding mode estimation is correct. In reality, however, erroneous coding mode estimation frequently occurs, ultimately resulting in poor performance. In the proposed method, we reduce the number of erroneously estimated coding-modes by utilizing the fact that MPEG-2 video coding is based on block-based coding. This enhanced hybrid technique effectively conceals the lost blocks by raising the probability of correct coding mode estimation by exploiting image characteristics such as scene changes and complex motions. Through computer simulations on damaged images we show that the proposed method has excellent performance.

This paper presents the implementation of MPEG-2 Advanced Audio Coding (AAC) decoder on a fixed-point DSP. AAC is a powerful coding algorithm and offers the high-quality multi-channel surround audio. We also discuss the implementation issue and effort in porting AAC algorithms to the DSP.

This paper presents a subband coding technique suitable for image compression in a single CCD camera with a Bayer color filter array (CFA). In it, we have applied a SSKF (symmetric short kernel filter) both horizontally and vertically to red and blue color signals, and a two dimensional perfect reconstruction filter to green color signals. Here, we compare this technique to two other image compression methods: DPCM and the Hadamard transform, each of which also allows an output signal from the CCD in a color camera to be compressed directly with simple logic circuitry and is suitable for use in low cost video conference cameras. Simulation results demonstrate that the subband coding offers the best quality (27-30 dB) with a compression ratio of approximately 2 bit/pel.

This article examines an original process to code signals from two different sources (data e.g. audio- and video) mixed together, and coded within a single coding process (DCT-based JPEG codec), in order to generate a single data stream at its output, instead of two as in the usual multimedia applications.

We present an MPEG decoder with reduced system costs by employing embedded compression of the reference frames which are used for motion compensated (MC) decoding. The embedded compression scheme used is based on either block predictive coding or DCT transform coding, depending on the required memory compression. The compression features simple recovery of (MC) block data, while preventing visible artifacts. The compression was optimized for low costs, enabling real application in a commercial MPEG IC. It was found that the same techniques can be applied in the encoder as well.

Speech coders based on linear predictive coding are good and accurate in modeling speech utterances. They not only model speech utterance in an accurate manner but have the following properties: i) they provide a good estimate of the vocal tract spectral envelope; ii) they are analytically tractable; iii) they can be implemented in either software or hardware; iv) they use less data storage than many other approaches of speech coders. In this paper, a hybrid speech coder based on code excited linear predictive coding (CELPC) and voice excited linear predictive coding (VELPC) is presented. CELPC is one of the main techniques for producing high quality speech at around 4.8 kbps. However, the computation requirement for the original CELPC was too demanding and not suitable for practical use. Therefore, a hybrid approach to produce speech signals with high quality and with a bit rate of 3.478 kbps has been proposed. This method splits the speech signal into two portions, the base-band signal and the high-band signal in the frequency domain. The base-band and the high-band signal are then coded using the CELPC and VELPC techniques, respectively.

In this paper, we propose a new wavelet-based image coding scheme which exploits jointly the statistical redundancy and the irrelevance properties of the human visual system. We first establish the visibility threshold model in the wavelet transform domain. Then, a visual compander is designed so that the output data of the compressor is visually uniform. A uniform quantizer and entropy coding are used for the output data. Experiment results show that our image coding scheme provides slightly improved visual quality without no degradation in PSNR, compared with the result of Shaprio's (1993) scheme.

This paper presents the implementation of MPEG-2 layer 1 audio coding and decoding system on a general purpose fixed-point DSP chip. A modified coding algorithm that alleviates the edge effect is described. This scheme encodes data in frames adjacent to the current frame so that data is closely correlated between frames. Results from the DSP real-time implementation are examined. Bit-rates from 74 to 138 kbps for the signals tested are attainable with transparent quality signal reconstruction.

The blocking effect is a common problem that is always present in image coding when using the blocking transform. A simple method for image coding, with low blocking effect, using pre- and post-axis rotation and the discrete sine transform, is proposed. It uses the overlap of only the sample and the total number of transform coefficients remains the same as the number of pixels in the original image. The method was implemented, and the results obtained by simulation in the image coding are compared with image coding by DCT, DST, and LOT.

We propose an efficient stereo audio coding scheme with error protection based on transformed vector quantization. This scheme includes a new stereo coding method, called adaptive inter-channel prediction stereo, that achieves high-efficiency stereo coding at a low coding bit rate. This scheme also includes an unequal error protection (UEP) method that achieves high-efficiency error protection against radio transmission error. We also report a real-time audio transfer system on the Japanese Personal Handyphone System (PHS) using the proposed coding scheme, which enables a high-quality audio transfer service over mobile channels.

In this paper, we propose a new method to reduce blocking effects in subband/wavelet image coding. Unlike the traditional postprocessing approaches, by using the block shifting method, the blocks in different bands are coded with different shifting positions. After subband/wavelet reconstruction, the block edges of different bands can be separated to reduce blocking effects. The simulation results are demonstrated to verify the proposed method.

This paper describes two novel methods to improve the efficiency of subband coding by means of the energy compactness in the frequency domain. Our coding scheme applies a variable block size DCT to the subband signal to get a high coding efficiency. A class driven segmentation is devised as a variable block partition scheme. Besides, an information oriented coding strategy without side information is developed. The strategy not only retains the image properties but also reduces the computation overhead. Simulations are carried out to the Lena image, and their results show the effectiveness of the two approaches especially for low bit rate applications.

A real-time coding that performs, in one coding pass, a variable bit rate coding that capitalizes on MPEG-2 video coding is proposed. In this method, coding is performed according to the degree of difficulty in image coding. To realize this objective, the relationship between quantization scale and generated bits is modeled as a bit amount assigning function. Based on this bit amount assigning function, the quantization scale is determined for each picture frame and coding is performed with this quantization scale. To verify this method, computer simulations are performed and MPEG-2 real-time encoders are built to realize the hardware installation of this method. From the results, it is confirmed that: 1. assignment of bit amounts in this method is executed in a similar fashion to that achieved in conventional two-pass variable bit rate coding, 2. Image quality can greatly be improved over the quality achieved in the constant bit rate coding, and 3. Coding performance remains stable even over extensive coding durations. Also, this method can be implemented using the same level of hardware as is required in constant bit rate coding. As a result, we have the prospect that two hours of high quality video can be recorded onto one side of a DVD-RAM disc in real time with the proposed method.

The image division coding is one of the means of coding the image signals with high resolution such as HDTV. Generally the parallel processing scheme is applied to carry out the image division coding by dividing the image in several panels spatially and then employing the coding tools for each panel. However, it is necessary to distinguish the difference in coding performance between parallel and non-parallel processing schemes. This paper deals with the buffer size and buffer delay for coding due to parallel and non-parallel processing schemes. The following conclusion has been deduced through such study. For the same rate control conditions, the J division parallel processing needs the buffer size J times larger than the non-parallel processing, with the buffer delay being the same in both cases.

In the MPEG 1 audio compression standard, the intensity stereo coding scheme is a key technology to achieve a bit rate lower than 2/spl times/128 kbit/s for stereophonic signals while preserving the high audio quality. However, the intensity coding scheme has difficulties in the signal cancellation and scale-factor consistency. This paper shows that the Karhunen-Loeve (KL) transform can be applied to the intensity coding for solving the two problems. We present two methods used to implement the Karhunen-Loeve transform in the MPEG 1 layers I and II. Among the two methods, one is fully compatible with the MPEG 1 while the other need a slight modification on the MPEG coding process. The two methods are tested and shown to have a better audio quality than the technique indicated by the MPEG 1 layers I and II.

MPEG-2 audio is the subband coding technology used for various audio applications. This paper presents a semi-custom ASIC design for the MPEG-2 audio decoder. The decoder implemented meets the requirements of the MPEG-2 international standard, and is divided into three parts: the preprocessor, the multichannel processor, and the synthesis filter. The decoder system has been designed in VHDL (VHSIC hardware description language), and developed as a single chip in a 0.6 /spl mu/m CMOS semiconductor process.

Subband coding of images has received considerable attention and has become an effective tool for image compression. In the work by Karlsson and Vetterli (1988), a 3-D subband coding of video using 2-tap symmetrical short-kernel filters for vertical filtering was suggested for transmission over packet-switched network environments. A modified short-kernel filter pair is proposed to eliminate the coding delay and to improve the reconstructed image quality. Simulations are conducted and satisfactory results are achieved. >

A VLSI architecture for a low complexity intraframe subband image coding for HDTV signals is presented. The generalized quadrature mirror filters (GQMFs), which have smaller overall delay, are optimized in order to achieve high coding efficiency. The filter design exploits a symmetry property among different filter coefficients which, in turn, reduces the hardware complexity of the VLSI architecture substantially. The architecture is designed with 1 /spl mu/m CMOS technology using the scalable design rules of MDS Integrated Services Inc. (MOSIS). The VLSI architecture contains approximately 200,000 transistors and it is capable of operating at a speed of 34 MHz. The HDTV signals are decomposed into seven bands. The baseband is DPCM encoded and the high bands are PCM encoded. It is shown that high quality HDTV images can be obtained at a low bit rate as 0.654 bpp (bits/pixel). >

3D video (3DV) data usually includes both conventional 2D videos and corresponding depth maps. In 3DV coding, color videos and depth maps need to be jointly coded. Usually, the system of 3DV coding uses the two-channel coding method which encodes color videos and depth maps based on two parallel codec implementations. The complexity and hardware requirements are nearly two times higher than coding 2D color videos. While low-complexity color video coding has been broadly studied, depth map coding has received much less attention. The depth map represents the distance from cameras to objects, which has characteristics of both data on Z-axis in world coordinate and the video signal. Thus, there is a high correlation among motion information (the prediction mode, reference frame and motion vector) from color videos and depth maps. Based on this observation, we propose one method to reduce depth coding complexity. An experimental analysis is performed to study the prediction mode correlation in coding information from color videos and depth maps. Based on the correlation, we propose an efficient mode decision algorithm. With almost the same RD performance, the proposed algorithm can reduce about 70% computational complexity of depth coding, which is beneficial for real-time realization through a hardware or software implementation for the 3DV applications.

A complete audio coding system which combines modified transform coding and dynamic bit allocation to achieve a two-channel, 15 kHz bandwidth, 256 kb/s data rate audio compression and decompression is proposed. The modified transformation is used for time-frequency representations of signals and the dynamic bit allocation scheme is used to progressively quantize the transform coefficients according to the rate distortion theory. Computer simulation results show that this system, which achieves high audio quality at the rate of 4 b/sample, possesses low computational complexity suitable for multimedia applications. This coding system can further be operated in the case of variable bit rate to allow application-specific ancillary data without drastic loss of signal quality. >

Precision adaptive subband coding is an audio coding method that allows a high-quality audio signal to be coded at a low bit rate. Basic design considerations, subjective testing for system improvement, a performance comparison, and the final implementation in integrated electronics are described. >

The design philosophy of the DCC is described. The DCC system is a digital extension of the well-known compact cassette system. The main differences are: highly efficient source coding of the audio signal, called precision adaptive subband coding (PASC), resulting in excellent sound quality; digital multitrack recording, including error correction techniques; thin film techniques for head production; autoreverse feature in all players; and new features (e.g., text display). The system performance, the cassette, the tape format, and the PASC are discussed. >

Video compression is achieved in the spectrum-compatible high-definition television (SC-HDTV) system by adaptive transform subband coding. The number of subbands is significantly increased, and the temporal processing is greatly reduced. Coefficients are transmitted as analog data; side information representing adaptation is transmitted as part of the digital data signal. >

The authors describe a single-chip, half-duplex multirate transcoder device. The device is pin selectable as a 32-kb/s ADPCM (adaptive differential pulse code modulation) transcoder and a 16-kb/s subband coding transcoder, and it provides a 64-kb/s PCM passthrough option. The transcoder implementation is fully compatible with the 32-kb/s ADPCM standard defined by ANSI and CCITT recommendation G.721 bis. The subband coder (SBC) algorithm provides a high-quality, cost-effective medium bit-rate solution. The ADPCM and SBC implementations on the device produce toll quality speech. >