This work proposes a novel formulation of convolutional-sparse-coded dynamic mode decomposition (CSC-DMD) incorporating a deep learning framework. CSC-DMD is a high-dimensional data analysis method with a convolutional synthesis dictionary and applicable to analyze dynamics such as seismic motions and river flows. An authors’ previous work has shown the effectiveness of CSC-DMD for riverbed state estimation. However, there still remains a room to improve the performance in expressing evolution of temporal and spatial changes in riverbed shape. Hence, this work proposes to adopt multi-resolution convolutional dictionary by introducing a deep learning framework so that the capability of simultaneously capturing local and global features is added to CSC-DMD. The significance of the proposed method is verified by evaluation of riverbed state estimation for time-series data of water surface and riverbed shape obtained through an experimental setup of river model.

This work proposes a novel image restoration technique inspired by the Ulyanov’s deep image prior (DIP) method. DIP uses a deep convolutional network as an image prior to generate a restored image from a random input one, which brings an advantage of no requirement of training data. However, one problem arises that the interpretability is low and it is not trivial to explain the need for the random input. This paper contributes to train a network by using the Stein’s unbiased risk estimator (SURE) with Monte-Carlo computation for self-supervised learning of image restoration. The Monte-Carlo calculation accepts a random input for computing divergence of the network, and allows us to interpret the reason why a random input is needed. Our framework trains an image restorer instead of an image generator, which simplifies the interpretation of the network structure. Thanks to this, the Luisier’s interscale linear expansion of the thresholding (LET) can become introduced to exploit the cross-correlation among extracted features. In order to avoid using group delay compensation and reduce the number of design parameters, a structured DIP is constructed by using non-separable oversampled lapped transform (NSOLT) instead of U-Net. By showing some results of denoising simulation, the significance of the proposed method is verified.

This study proposes a filtering method for directed graph (digraph) signals. In order to realize digraph filtering, a novel graph Fourier transform (GFT), – Augmented GFT (AuGFT) –, is proposed by defining an Hermitian adjacency matrix. Although there has been the same method to give the adjacency matrix of digraphs, this study defines a novel digraph Laplacian. The existing digraph Laplacian does not give the graph signal variation considering the edge directions, while the novel one does. This paper introduces three important ideas. The first is the definition of a novel degree matrix to give the novel digraph Laplacian. The second is to decompose the symmetric and skew-symmetric components of the novel digraph Laplacian independently into their spectral components. The third is, based on the decomposition, to augment the conventional GFT for digraphs as an invertible real-valued dictionary. The new GFT is shown to provide a practical form of real-valued digraph filtering. The significance of the proposed method is verified through simulations of signal filtering on digraphs.

This study extends a self-supervised image denoising technique proposed by the authors to a more general image restoration method. The previous work was inspired by Ulyanov’s deep image prior (DIP) method. DIP uses a deep convolutional network as an image prior to generate a restored image from a random one, which brings an advantage of no training data requirement. However, one problem arises that the interpretability is low and it is not trivial to explain the need for the random input. In the previous work, it is shown that Stein’s unbiased risk estimator (SURE) with Monte-Carlo computation can be used to train a denoising network. This approach allows us to interpret the random input requirement. The framework trains an image denoiser instead of an image generator. This work extends the discussion to more general image restoration problem by introducing a deep unrolling approach to reflect the measurement process. The new framework uses the Luisier’s inter-scale linear expansion of the thresholding (LET) for exploiting the cross-correlation among extracted features. Some simulation results of image restoration show the significance of the proposed method.

This study proposes a novel approach for the analysis of high-dimensional time series data utilizing a linear shift-variant transform. Central to the methodology is the Locally-Structured Unitary Network (LSUN), a self-supervised learning transform proficient in capturing the underlying tangent spaces within high-dimensional datasets. This trait equips LSUN with a distinct advantage over existing techniques such as convolutional dictionary learning, which suffers from decreased interpretability due to the redundant selection of filter kernels. The shift-invariance of convolutional models does not directly correspond to the axes of tangent spaces, a challenge well addressed by LSUN. This study focuses on utilizing LSUN for time series analysis to capture structural variations in sequential frames. To optimize this task, LSUN is utilized with total variation regularization, a method known for its ability to preserve edge information. The effectiveness of the proposed model is demonstrated through a video sequence denoising experiment, illustrating its potential to innovate in areas where the analysis of time-series data is crucial. The findings signal the significant potential for this approach to be extended to targets, such as dynamical system identification, further extending its applicability.

This study proposes a flow-path fitting method to asses river health condition. In recent years, river flooding due to abnormal weather has been a growing problem in many parts of the world. Meandering of rivers is one of the causes of river flooding. In order to solve this problem, the authors have proposed a river flow path control cyber-physical system (CPS). The CPS adopts reinforcement learning (RL) to control actuators that act as groyens. To realize the RL, a reward is needed to index the river health. First, this paper defines a river health index on the assumption that the flow path is represented by a function, and evaluates its energy. However, it is not trivial to identify the dominant path from river videos captured by cameras or radars due to false detections, undetections and noise. In order to obtain a dominant path by image processing, this study reduces the problem to a group LASSO one using Fourier basis for unequally spaced and repeatedly sampled noisy data. The solver is given by ADMM. The significance of the proposed method is verified by evaluating its performance through simulations using artificial data and experiments using a river model setup.

This work proposes a method for estimating dynamics on graph by using dynamic mode decomposition (DMD) and sparse approximation with graph filter banks (GFBs). The motivation of introducing DMD on graph is to predict multi-point river water levels for forecasting river flood and giving proper evacuation warnings. The proposed method represents a spatio-temporal variation of physical quantities on a graph as a time-evolution equation. Specifically, water level observation data available on the Internet is collected by web scraping. As well, the graph structure is defined based on numerical river information published by Ministry of Land, Infrastructure, Transport and Tourism (MILT) of Japan and the graph is used to construct GFBs for analyzing and synthesizing the water level data. GFBs work in combination with a sparse approximation algorithm for feature extraction of water level distribution. The features are exploited to derive the time-evolution equation through the extended DMD (EDMD) framework. The time-evolution equation is applied to predict river water level distribution. In order to verify the significance of the proposed method, the river water level prediction is conducted for real web-scraped data. The performance evaluation shows the superiority to the normal DMD approach.

This work proposes convolutional-sparse-coded dynamic mode decomposition (CSC-DMD) by unifying extended dynamic mode decomposition (EDMD) and convolutional sparse coding. EDMD is a data-driven method of analysis used to describe a nonlinear dynamical system with a linear time-evolution equation. Compared with existing EDMD methods, CSC-DMD has the advantage of reflecting the spatial structure of a target. As an example, the proposed method is applied to river bed shape estimation from the water surface observation. This estimation problem is reduced to sparsityaware signal restoration with a hard constraint given by the CSC-DMD prediction, where the algorithm is derived by the primal-dual splitting method. A time series set of water surface and bed shape measured through an experimental river setup is used to train and test the system. From the result, the efficacy of the proposed method is verified.

This work proposes a novel restoration model for optical coherence tomography (OCT) data. The authors have been developing a multi-frequency swept (MS) en-face OCT device that can help understand the mechanism of the sensory epithelium in the cochlear. Although the device has merit in acquiring moving tissues, the broadened light gives a weak response; thus, some signal restorations are demanded. This work proposes the introduction of a formulation for OCT data restoration as a convex optimization problem by assuming a latent refractive index distribution. An algorithm to solve the problem with the primal-dual splitting (PDS) framework is then derived. The PDS has an advantage of requiring no inverse matrix operation and being able to handle high-dimensional data. The significance of the proposed model is verified by simulations on artificial data, followed by an experiment with the actual observation of 256 256 2000 voxels.

This work proposes a volumetric data restoration method, especially for data acquired through an optical coherence tomography (OCT) device. OCT is a technique for acquiring a tomographic image of a specimen object in a few μm scale by using a near infrared laser. The authors have been trying dynamic observation of epithelium in cochlear of the inner ear. Currently, there is a problem to remove the influence of the measurement process as well as noise due to image sensor sensitivity. Therefore, in this work, on the assumption that specimen objects follow some sort of signal generation model, an OCT volumetric data restoration method is proposed. The proposed technique adopts the primal-dual plug-and-play (PDPnP) method, where the generation model is represented by a sparsity-aware regularization term explicitly or implicitly. The significance of the proposed method is verified by simulation on artificial data, followed by an experiment with actual observation data.

Redundant-transform-based image fusion approaches require a lot of memory and computations. This paper proposes an effective and efficient multi-focus image fusion technique in dual domain using a redundant transform. The proposed scheme captures high-frequency informations, e.g. edges and slant textures, of images efficiently, and reduce the computational cost. The proposed scheme extracts the high-frequency information of images with multiple directional lapped orthogonal transforms (M-DirLOTs) through the following procedure: (1) reconstruct the detail image using high-pass subbands, (2) execute a fusion operation in spatial domain through joint measurement and (3) improve the performance by mathematical morphology processing. The proposed method overcomes some disadvantages of traditional transform-based and spatial-based fusion techniques. Experimental results show that the proposed method is able to significantly improve the fusion performance.

Recently, single image super-resolution (SISR) is very important research field to reconstruct a high-resolution (HR) image from a low-resolution (LR) image. However, existing image super-resolution approaches require a lot of computations or consider parameters for various situations. This paper proposes an efficient and simple image super-resolution technique using multiple directional lapped orthogonal transforms (M-DirLOTs). It captures high-frequency informations, e.g. edges and slant textures, of images efficiently, and reduce the computational cost. Simultaneously, this model avoids any a priori hypotheses on the LR picture. The proposed method overcomes some disadvantages of existing methods. Experimental results show that the proposed method is able to significantly improve the superresolution performance.

This paper proposes an efficient design method of nonseparable oversampled lapped transform (NSOLT). NSOLT is a multidimensional redundant transform which satisfies the nonseparable, symmetric, real-valued, overlapped, compact-supported and perfect-reconstruction property. A typical example-based design approach, which consists of sparse coding and parameter optimization, is applicable to NSOLT. In the previous implementation, however, the parameter optimization stage dominated the computation. The main reason is that the quasi-Newton method with numerical gradient of the objective function was adopted. To reduce the computational cost, the analytical gradient is derived and introduced. For further acceleration, the quasi-Newton method is replaced by stochastic gradient descent. Through some experiments, the significance of the proposed method is verified.

This paper proposes a Poisson denoising with a union of directional lapped orthogonal transforms (DirLOTs). DirLOTs are 2-D non-separable lapped orthogonal transforms with directional characteristics. Its bases overcome a disadvantage of the separable wavelet image denoising for the diagonal textures and edges. Based on this feature, multiple DirLOTs are used to improve the performance by introducing redundant representation with multiple directions. Experimental results show the combination of the variance stabilizing transformation (VST), Stein's unbiased risk estimator-linear expansion of thresholds (SURE-LET) approach and multiple DirLOTs is able to significantly improve the denoising performance, and verify the feasibility of the proposed method.

This work proposes a novel design method of a two-dimensional (2-D) Non-Separable Oversampled Lapped Transform (NSOLT) for a given image by introducing a typical two stage procedure of dictionary learning. NSOLT is a lattice-structure-based transform and yields a redundant dictionary of which atoms satisfy the non-separable, symmetric, real-valued, overlapping and compact-support property. In addition, the Parseval tight frame constraint can structurally be imposed, while the redundancy R is flexibly controlled by the ratio of the number of channels P and the downsampling ratio M. Compared with the other dictionary learning approaches, the proposed method is moderately structured so that it is capable of multiscale construction as well as atom termination at image boundary. The significance of the proposed method is verified by showing an example of learned dictionary and sparse approximation results.

This paper proposes lattice structures for two-dimensional (2-D) non-separable (NS) oversampled (OS) lapped transforms. The proposed systems consist of the 2-D separable discrete cosine transform and NS support extension processes, and allow us to simply realize rational redundancy as well as the overlapping, paraunitary (PU), symmetric, real-valued and compact support property. The lattice structures have two aspects. One is an extention of OS linear-phase (LP) perfect reconstruction (PR) filter banks (FBs) to the 2-D NS case. The other is a generalization of 2-D NS LPPR FBs to the OS case. The significance is verified by showing some design examples and image restoration results with the iterative shrinkage/ thresholding algorithm (ISTA).

A color-tone similarity index (CSIM) between two color images is presented. CSIM is defined by a statistical analysis of cumulative histograms in a hue-oriented color space. It characterizes the color distributions, while the existing structural similarity index reflects the spatial structure involved with grayscale images. The behaviors of CSIM are checked by the comparisons of color code chips. Through an image quality assessment on TID2008, the correlation between CSIM and the mean opinion score was proved to be statistically significant.

This work proposes to apply a two-dimensional (2-D) non-separable oversampled lapped transform (NSOLT) to image restoration. NSOLT is a lattice-structure-based redundant transform which satisfies the symmetric, real-valued and compact-support property. The lattice structure is able to constitute a Parseval frame with rational redundancy and produce a dictionary with directional atoms. In this study, the performance for deblurring, super-resolution and inpainting is evaluated. The iterative-shrinkage/thresholding algorithm (ISTA) is adopted to show the significance of NSOLT in the image restoration applications. It is verified that the six-level NSOLT with redundancy less than two yields superior or comparable restoration performance to the two-level non-subsampled Haar transform of redundancy seven in both of PSNR and SSIM.

A novel image denoising technique is proposed by using directional lapped orthogonal transforms (DirLOTs). DirLOTs satisfy orthogonality and the bases are allowed to be anisotropic with the fixed-critically-subsampling, overlapping, symmetric, real-valued and compact-support property. In this work, DirLOTs are used to construct directional symmetric orthonormal discrete wavelet transforms and then the bases are adopted to generate a redundant dictionary with several directions. The multiple directional property is suitable for representing natural images which contain diagonal edges and textures. The proposed dictionary is applied to solve the basis pursuit denoising problem. The denoising performance is evaluated for several images through the heuristic shrinkage and block-coordinate-relaxation algorithm. It is verified that the proposed technique is simple but yields perceptually preferable results.

This paper proposes to adopt a union of directional lapped orthogonal transforms (DirLOTs) to the SURE-LET image denoising. DirLOTs are 2-D non-separable transforms and can have directional characteristics under the fixed-critically-subsampling, overlapping, orthogonal, symmetric, real-valued and compact-support property. As a previous work, the author showed that a DirLOT overcomes a disadvantage of separable wavelet image denoising for diagonal textures and edges. Since there remains a problem that the approach is applicable only to a single geometric direction, this work further investigates to improve the performance by introducing redundant representation with multiple directions. Through some experimental results, it is verified that the proposed technique shows better performance than the traditional single wavelet approach for natural images with rich amount of geometrical structures.

This paper proposes a directional design method of 2-D nonseparable linear-phase paraunitary filter banks. The proposed method is based on a lattice structure consisting of the 2-D separable DCT block and nonseparable support extension processes. Because of the nonseparability, the bases are allowed to be directional with the critically fixed subsampling, overlapping, orthogonal, symmetric, real-valued, and compact support properties. First, a novel vanishing moment (VM) condition is introduced as a suitable directional constraint, where the moment is referred to as the trend VM. The condition forces wavelet filters, i.e., high-pass and bandpass filters, to annihilate trend-surface components. Second, some theoretical properties of TVMs are discussed for general 2-D paraunitary systems, and then, the properties are applied to the lattice parameters. In order to verify the significance, several design examples are shown, the trend-surface annihilation properties are numerically confirmed, and the denoising capability is evaluated for images through shrinkage. It is shown that our proposed transforms yield perceptually preferable results.

This paper proposes a boundary operation technique of 2-D nonseparable linear-phase paraunitary filter banks (NS-LPPUFBs) for size limitation. The proposed technique is based on a lattice structure consisting of the 2-D separable block discrete cosine transform and nonseparable support-extension processes. The bases are allowed to be anisotropic with the fixed critically subsampling, overlapping, orthogonal, symmetric, real-valued, and compact-support properties. First, the blockwise implementation is developed so that the basis images can be locally controlled. The local control of basis images is shown to maintain orthogonality. This property leads a basis termination (BT) technique as a boundary operation. The technique overcomes the drawback of NS-LPPUFBs that the popular symmetric extension method is invalid. Through some experimental results of diagonal texture coding, the significance of the BT is verified.

A novel design method of 2-D non-separable linear-phase paraunitary filter banks (LPPUFBs) is proposed. The trend vanishing moment (TVM) condition is newly derived for the lattice structure. The TVM condition can be regarded as a natural extension of 1-D VM to 2-D one and alternative of the conventional directional vanishing moments(DVMs). The structure is based on the 2-D DCT and can be regarded as an extension of GenLOT to the 2-D non-separable case. Because of the non-separability, the bases are allowed to be anisotropic and directional in addition to the fixed-subsampling, overlapping, orthogonal, real-valued, symmetric and compact-support property. As a previous work, the authors have shown that the block-wise implementation is appealing for the image processing application since the technique serves boundary operation for size-limitation, seamless basis alteration and compatibility with the block DCT. This work further contributes to yield the TVM condition so that piece-wise smooth images with smooth contours are compactly approximated. The condition is imposed directly on the design parameters and, therefore, the desirable properties of the lattice structure are maintained. From an experimental result of coding, it is shown that the TVM condition gives smooth reconstruction of a diagonal texture image.

This work proposes a low complexity computation of EM algorithm for Gaussian mixture model(GMM) and accelerates the parameter estimation. In previous works, the authors revealed that the computational complexity of GMM-based classification can be reduced by using an interval calculation technique. This work applies the idea to EM algorithm for GMM parameter estimation. From experiments, it is confirmed that the computational speed of the proposal achieves more than twice that of the standard method with 'exp( )' function. The relative errors are less than 0.6% and 0.053% when the number of bits for table addressing are 4 and 8, respectively.

A scaled transform is defined for a given irreversible linear transformation based on the LU factorization of a nonsingular matrix so that the transformation may be computed in a lifting form and hence may be reversible. Round-off errors in the lifting computation and the computational complexity are analyzed. Some reversible component transforms are presented and experimented to give some remarks to image compression applications. Discussions are developed with coding gain and actual bit rates.

This work contributes to investigate theoretical properties of the trend vanishing moment (TVM) which the authors have defined in a previous work and applied to the directional design of 2-D nonsep-arable GenLOTs. The TVM condition can be regarded as a natural extention of 1-D VM to 2-D one and alternative of the conventional directional vanishing moment (DVM). Firstly, the definition of TVM is given to clarify what it is, and then the mathematical meaning is discussed. While the conventional DVM condition requires for the moments to vanish along lines in the frequency domain and restricts the direction to a rational factor, the TVM condition imposes the moments only point-wisely and the direction can be steered flexibly. Some significant properties of TVMs are also shown and a simulation result of zonal coding for an artifical picture is given to verify the trend surface annihilation property.

This work proposes a fast decision algorithm in pattern classification based on Gaussian mixture models (GMM). Statistical pattern classification problems often meet a situation that comparison between probabilities is obvious and involve redundant computations. When GMM is adopted for the probability model, the exponential function should be evaluated. This work firstly reduces the exponential computations to simple and rough interval calculations. The exponential function is realized by scaling and multiplication with powers of two so that the decision is efficiently realized. For finer decision, a refinement process is also proposed. In order to verify the significance, experimental results on TI DM6437 EVM board are shown through the application to a skin-color extraction problem. It is verified that the classification was almost completed without any refinement process and the refinement process can proceed the residual decisions.

This paper introduces block-wise implementation of 2D non-separable linear-phase paraunitary filter banks (LPPUFB) and shows a directional design approach. As a previous work, the authors have proposed a lattice structure of non-separable LPPUFBs which guarantees both of the linear-phase and orthogonal property. This paper shows that the structure serves variability of basis images without any violation to the orthogonality. Consequently, a boundary operation for size-limitation is yielded and compatibility with the 2D separable block-DCT becomes possible since the lattice structure is based on the block-DCT. The main advantage of non-separable systems is its directional capability, and it is allowed to produce a directional non-separable GenLOT. Although the adaptive control of basis images is under investigation, experimental results of texture coding show prospective significance of the directional GenLOT.

A simple lossy-to-lossless bit-plane coding of still images is presented to integrate several functionality extension including selective tile partitioning, progressive transmission, ROI transmission, accuracy scalability, and others. The mean squared error between the original image and a decoded image at any progression level is known prior to encoding/decoding. The proposed bit-plane codec is competitive with JPEG-LS and JPEG 2000 in the lossless compression of 8-bit grayscale and 24-bit color images. The codec outperforms the existing standards in 8-bit color-quantized image compression.

This paper proposes a new type of layered video coding especially for the use of monitoring a river or a water channel. A sensor node of the system decomposes a video signal into some components and produces a bit stream which is functionally separated into three layers. The first layer contains the minimum components effective for detecting water level. The second layer contains signals for thumb-nail video browsing. Each of them is transmitted at very low bit rate for regular monitoring. The third layer contains additional data for decoding the original video signal. It is transmitted in case of necessity. The original video signal is decomposed into band signals as the components by the Haar transform in a sensor node. Experimental results show which band signals should be included into the first layer considering both of water level detection performance and data size to be transmitted.

In this work, a novel motion compensated spatio-temporal filter (MCSTF) is proposed for scalable video coding. MCSTF is an alternative technique of existing motion-compensated temporal filtering (MCTF), a fundamental component for scalable video coding in next generation. MCSTF is a non-separable sub-sampling version of MCTF and provides interlaced pictures as intermediate video sequence by using a spatio-temporal split process. Furthermore, the 1/3-transform structure, which excludes the lifting-update-step, significantly suppresses the PSNR fluctuation which occurs in the existing MCTF technique. Our proposed system has an advantage of suppressing PSNR fluctuation with almost the same average PSNR as that of existing MCTF. In this paper, two types of sub-sampling lattices are investigated: the vertical-temporal (VT) quincunx and face-centered orthorhombic (FCO) lattices. Some experimental results with entropy coded scalar quantization show the significance of our proposed technique.

This paper proposes a method of lossless image coding by the aid of lossy image coding. It aims at an improvement in the compression efficiency. We apply a kind of embedded coding to large coefficients in magnitude in a wavelet transform domain. The other wavelet coefficients are encoded by a context-based entropy coding. The result slightly outperforms the compression efficiency in JPEG-LS.

This paper presents an implementation scheme of Motion-JPEG2000 (MJP2) integrated with invertible deinterlacing. In previous work, we developed an invertible deinterlacing technique that suppresses the comb-tooth artifacts which are caused by field interleaving for interlaced scanning videos, and affect the quality of scalable frame-based codecs, such as MJP2. Our technique has two features, where sampling density is preserved and image quality is recovered by an inverse process. When no codec is placed between the deinterlacer and inverse process, the original video is perfectly reconstructed. Otherwise, it is almost completely recovered. We suggest an application scenario of this invertible deinterlacer for enhancing the sophisticated signal-to-noise ratio scalability in the frame-based MJP2 coding. The proposed system suppresses the comb-tooth artifacts at low bitrates, while enabling the quality recovery through its inverse process at high bitrates within the standard bitstream format. The main purpose of this paper is to present a system that yields high quality recovery for an MJP2 codec. We demonstrate that our invertible deinterlacer can be embedded into the discrete.wavelet transform employed in MJP2. As a result, the energy gain factor to control rate-distortion characteristics can be compensated for optimal compression. Simulation results show that the recovery of quality is improved by, for example, more than 2.0 dB in peak signal-to-noise ratio by applying our proposed gain compensation when decoding 8-bit grayscale Football sequence at 2.0 bpp.

Perfect reconstruction (PR) deinterlacer banks are proposed as new tools for spatio-temporal scalable video codec. The proposed systems separate a progressive video into two different progressive videos of a half frame rate and are novel from the viewpoint of filter-banks in that interlaced videos are given as intermediate data during analysis and synthesis process. Unlike the conventional filter banks, our systems are constructed in a way unique to multidimensional systems by using invertible deinterlacers which we have proposed before. This paper suggests two kinds of interlaced video formats: the vertical-temporal (VT) Quincunx and face-centered-orthorhombic (FCO) format. As a primal application, a spatio-temporal scalable video codec, or field scalable video codec (FSVC), system is experimented. This technique offers a functionality by which decoding a base layer by itself provides both of half-rate progressive and interlaced videos, and adding an enhancement layer improves the spatio-temporal resolution. Some experimental results of the filter banks combined with JPEG2000 show their potential for practical video codec applications.

In this paper, we derive spatial correlation functions of linear and circular antenna arrays for three types of angular energy distributions: a Gaussian angle distribution, the angular energy distribution arising from a Gaussian spatial distribution, and uniform angular distribution. The spatial correlation functions are investigated carefully. The spatial correlation is a function of antenna spacing, array geometry and the angular energy distribution. In order to emphasize the research and their applications in diversity reception, as an example, performance of the antenna arrays with MRC in correlated Nakagami fading channels is investigated, in which analytical formulas of average BER for the spatial correlation are obtained.

Invertible deinterlacing with variable coefficients is proposed to suppress comb-tooth artifacts caused by field interleaving of interlaced scanning video. A vertical highpass filter is applied to detect moving artifacts around boundaries of moving objects. The coefficients of a deinterlacing filter is varied depending on the motion intensity so that the deinterlacing filter may be matched to the local characteristics of moving pictures. Note that the deinterlacing filter is motion-adaptive and is time/translation-varying, while the deinterlacing is still kept to be invertible. The deinterlacing filter performance and its contribution to intraframe-based video coding are evaluated. In addition, since the processing of motion detection and a part of deinterlacing filtering can be shared, their efficient implementation is derived in the form of lifting popular in wavelets.

In this work, discrete-wavelet-transform (DWT) gain compensation of Motion-JPEG2000 (MJP2) for invertible deinterlacer is proposed. As previous works, we have developed invertible deinterlacer that suppresses comb-tooth artifacts caused by field interleaving for interlaced scanning video, which affect the quality of scalable intraframe-based codec such as MJP2. Our technique has two features that the sampling density is preserved and the image quality can be recovered by the inverse process on demand. When no codec is applied in between the deinterlacer and inverse process, the original video is perfectly reconstructed. Otherwise, it is approximately recovered. The purpose of this work is to improve the quality of recovered images when MJP2 codec is inserted. It is shown that our invertible deinterlacer can be embedded into the DWT. As a result, frequency weighting for rate-distortion control can be moderately compensated. Simulation results show that the quality recovery is improved more than 1.5 dB in PSNR by applying the proposed compensation compared with the original weighting at 2.0 bpp of decoding rate for 8-bit grayscale pictures.

Invertible deinterlacing with variable coefficients is proposed to suppress comb-tooth artifacts caused by field interleaving of interlaced scanning video. A vertical highpass filter is applied to detect moving artifacts around boundaries of moving objects. The coefficients of a deinterlacing filter are varied depending on the motion intensity so that the deinterlacing filter may be matched to the local characteristics of moving pictures. Note that the deinterlacing filter is motion-adaptive and is time/translation-varying, while the deinterlacing is still kept to be invertible. The deinterlacing filter performance and its contribution to intraframe-based video coding is evaluated. In addition, since the processing of motion detection and a part of deinterlacing filtering can be shared, their efficient implementation is derived in the form of lifting popular in wavelets.

In this paper, a simple and accurate online SNR estimator is proposed for parallel combinatorial SS (PC/SS) systems in Nakagami fading channels. The PC/SS systems are a kind of partial-code-parallel multicode DS/SS systems. Our proposed SNR estimator estimates the SNR at the correlator output by using a statistical ratio of the sum of correlator outputs for pre-assigned PN codes. We investigate the SNR estimation accuracy in Nakagami fading channels through computer simulations. Numerical results show that the the proposed SNR estimator brings superior estimation performance to that in conventional estimator that was proposed by Summers and Wilson (1998). In addition, we apply it to the convolutional coded PC/SS systems with iterative demodulation and decoding to evaluate the estimation performance from the viewpoint of error rate. By using our estimator, the bit error rate performance is close to the performance with perfect knowledge of the SNR information in Nakagami fading channels and correlated Rayleigh fading channels.

A new class of de-interlacing is developed for intra-frame-based motion picture coding such as Motion-JPEG2000. The proposed technique has two features: the sampling density preservation and the invertibility. Thus, the amount of de-interlaced pictures is not increased and the original pictures can perfectly be reconstructed. This technique is regarded as a generalization of the conventional field interleaving. The significance is, verified by showing some design examples and simulation results, The proposed technique is also shown to cause less comb-shape artifacts than the conventional one. The intra-field-based case is also discussed.

We propose a forward error correction (FEC) coding scheme for partial-code-parallel multicode direct sequence spread spectrum (DS/SS) systems referred to as parallel combinatorial spread spectrum (PC/SS) systems. Information data are conveyed in two ways: the combination of orthogonal codes and the polarity or phase of their transmitting orthogonal code. It is known that FEC coding is an essential factor in the SS systems, especially in the CDMA environment. Straightforward FEC coding shows significant performance improvements. However, it could reduce the data transmission rate of PC/SS systems. From the viewpoint of error rate performance in the PC/SS systems, estimation of a combination of orthogonal codes dominates the total symbol and bit error rate performance. In this paper, FEC coding is separately applied to the combination data of orthogonal codes and the phase data of their transmitting orthogonal code, and less redundant FEC code is used. Therefore, the total information rate exceeds that offered by the conventional FEC coding. Simulation results show that the error rate performance of the PC/SS systems equipped with the separate FEC coding is lower than it is when conventional FEC coding is applied to the PC/SS system, in spite of having higher overall coding rate.

In this paper, we propose an approach to solving the power control issue in a DS-CDMA cellular system using genetic algorithms (GA). The transmitter power control developed in this paper has been proven to be efficient to control co-channel interference, to increase bandwidth utilization and to balance the comprehensive services that are shared among all the mobiles with attaining a common signal-to-interference ratio (SIR). In this paper, the optimal centralized power control (CPC) vector is characterized and its optimal solution for CPC is presented using GA, in which first in-first out (FIFO) stacks and nonlinear decreasing functions are derived in the investigation for enforcing the convergence rate. Emphasis is put on the balance of services and convergence rate by using GA.

This paper proposes iterative demodulation/decoding for parallel combinatorial spread spectrum (PC/SS) systems. The PC/SS systems convey the information data by a combination of pre-assigned spreading sequences with polarity. Convolutional coding with a random interleaver is implemented as channel coding, like serial concatenated coding. A 'soft-in/soft-out' PC/SS demodulator, based on an a posteriori probability algorithm, is proposed to perform the iterative demodulation and decoding. Simulation results demonstrate that the proposed iterative demodulation and decoding bring significant improvement in the bit error rate performance.

In the MPEG standard, motion estimation (ME) is used to eliminate the temporal redundancy of video frames. This ME is the most time-consuming task in the encoding of video sequences and is also the one using the most power. Using low-bit images can save the power of ME and a conventional architecture fixed to a certain bit width is used for low-bit motion estimation. It is known that there is a trade-off between power and image quality. ME may be used in various situations, and the relation between demands for power or image quality will depend on those circumstances. We therefore develop an architecture for a low-bit motion estimator with adjustable power consumption. In this architecture, we can select the bit width for the input image and adjust the amount of power for ME. To evaluate its effectiveness, we designed the motion estimator by VHDL and used the synthesis results to estimate the performance.

In this paper, we propose two multidimensional parallel processing methods for rational sampling lattice alteration. Our proposed methods enable us both to implement the rational lattice alteration with the parallel processor approach and to eliminate the redundancy caused by up- and down-sampling. Those methods are provided by extending two conventional block processing techniques for FIR filtering: the overlap-add method and the overlap-save method, respectively.

The overlap-add method (OLA) and overlap-save method (OLS) are well known as efficient schemes for high-order FIR filtering. In this paper, new sampling rate conversion methods are proposed by extending the OLA and OLS, and eliminating the redundancy caused by the conversion. First, for finite-duration sequences, a rate conversion with the DFT-domain approach is discussed. Next, using the result, the extended OLA and OLS are proposed for infinite-duration sequences. Finally, the computation complexities of our proposed methods are shown.>