This paper proposes an object-based approach to a class of dynamic image-based representations called ldquoplenoptic videos,rdquo where the plenoptic video sequences are segmented into image-based rendering (IBR) objects each with its image sequence, depth map, and other relevant information such as shape and alpha information. This allows desirable functionalities such as scalability of contents, error resilience, and interactivity with individual IBR objects to be supported. Moreover, the rendering quality in scenes with large depth variations can also be improved considerably. A portable capturing system consisting of two linear camera arrays was developed to verify the proposed approach. An important step in the object-based approach is to segment the objects in video streams into layers or IBR objects. To reduce the time for segmenting plenoptic videos under the semiautomatic technique, a new object tracking method based on the level-set method is proposed. Due to possible segmentation errors around object boundaries, natural matting with Bayesian approach is also incorporated into our system. Furthermore, extensions of conventional image processing algorithms to these IBR objects are studied and illustrated with examples. Experimental results are given to illustrate the efficiency of the tracking, matting, rendering, and processing algorithms under the proposed object-based framework.

Recent attention has been drawn to the hybrid Delta-Sigma (ΔΣ) structure featuring the integration of continuous-time (CT) and discrete-time (DT) structures in the loop filter. It combines the accurate loop filter characteristic of a DT ΔΣ modulator and the inherent anti-aliasing of a CT modulator. We present a design methodology for building a CT-DT ΔΣ modulator via the transformation from a DT ΔΣ modulator prototype. We also demonstrate the tradeoff of applying this structure to cascaded Delta-Sigma modulators compared to pure CT or DT implementations.

This paper proposes a new recursive adaptive beamforming algorithms for uniform concentric spherical array (UCSA) having nearly frequency invariant (FI) characteristics. New recursive adaptive beamforming algorithms based on the least mean square (LMS) algorithm and the generalized sidelobe canceller (GSC) structure are proposed. Simulation results show that the proposed adaptive FI-UCSA beamformer requires much fewer variable taps than the conventional UCSA for the same steady-state performance, while offering much faster convergence speed. Simulation results also show that the proposed FI-UCSA has uniform resolutions around both the elevation and azimuth angles.

This letter proposes a new method for designing finite-impulse response (FIR) filters with variable characteristics. The impulse response of the variable digital filter (VDF) is parameterized as a linear combination of functions in the spectral or tuning parameters. Using the least square objective function, the optimal solution is obtained by solving a system of linear equations. Design results show that this method is simple and effective in designing FIR VDF with good frequency characteristics. Furthermore, by using a piecewise polynomial, instead of an ordinary polynomial, more complicated frequency characteristics, or a larger tuning range can be approximated.

This paper proposes a new method for designing a class of two-channel perfect reconstruction (PR) linear-phase FIR filterbanks (FBs) and wavelets previously proposed by Phoong et al. By expressing the given K-regularity constraints as a set of linear equality constraints in the design variables, the design problem using the minimax error criterion can be solved using semidefinite programming (SDP). Design examples show that the proposed method is very effective and it yields equiripple stopband response while satisfying the given K-regularity condition.

This article proposes an efficient implementation of the Farrow (1988) structure using sum-of-powers-of-two (SOPOT) coefficients and multiplier-block (MB). In particular, a novel algorithm for designing the Farrow coefficients in SOPOT form is detailed. Using the SOPOT coefficient representation, coefficient multiplication can be implemented with limited number of shifts and additions. Using MB, the redundancy between multipliers can be fully exploited through the reuse of the intermediate results generated. Design examples show that the proposed method can greatly reduce the complexity of the Farrow structure while providing comparable phase and amplitude responses.

This paper proposes new integer approximations of the lapped transforms, called the integer lapped transforms (ILT), and studies their applications to image coding. The ILT are derived from a set of orthogonal sinusoidal transforms having short integer coefficients, which can be implemented with simple integer arithmetic. By employing the same scaling constants in these integer sinusoidal transforms, integer versions of the lapped orthogonal transform (LOT), the lapped biorthogonal transform (LBT), and the hierarchical lapped biorthogonal transform (HLBT) are developed. The ILTs with 5-b integer coefficients are found to have similar coding gain (within 0.06 dB) and image coding performances as their real-valued counterparts. Furthermore, by representing these integer coefficients as sum of powers-of-two coefficients (SOPOT), multiplier-less lapped transforms with very low implementation complexity are obtained. In particular, the implementation of the eight-channel multiplier-less integer LOT (ILOT), LBT (ILBT), and HLBT (IHLBT) require 90 additions and 44 shifts, 98 additions and 59 shifts, and 70 additions and 38 shifts, respectively.

Subspace tracking is an efficient method to reduce the complexity of signal subspace estimation. Recursive least square-based (RLS) subspace tracking algorithms such as the PAST algorithm is attractive because they estimate the signal subspace adaptively and continuously and the computational complexity is relatively low. Unfortunately, the RLS algorithm is well known to be very sensitive to impulse noise and it's performance can degraded substantially. In this paper, a robust PAST algorithm, based on the concept of robust statistics, is proposed. The robustness is achieved by making the underlying RLS iteration more robust to impulse interference. This new method is also applicable to other RLS-based algorithms. In particular, a robust statistic based impulsive noise detector is incorporated into the subspace tracking algorithm. The impulses in the input data vector are detected, and they are prevented from corrupting the estimated subspace for further tracking. We also propose a new restoring mechanism to handle long burst of consecutive impulses, which is a very difficult problem to handle in practice. Simulation results show the proposed algorithm offers satisfactory robustness against individual and consecutive impulses, while the PAST algorithm degrades dramatically in similar impulse noise environment. For nominal Gaussian noise, the proposed robust subspace tracking algorithm offers similar performance as the PAST algorithm.

This paper proposes a new design and implementation method for a class of multiplierless 2-channel 1D and 2D nonseparable perfect reconstruction (PR) filterbanks (FB). It is based on the structure proposed by S.M. Phoong et al. (see IEEE Trans. Sig. Proc., vol.43, p.649-64, 1995) and the use of multiplier blocks (MB). The latter technique allows one to further reduce the number of adders in implementing these multiplier-less FB by almost 50%, compared to the conventional method using sum of powers of two coefficients (SOPOT) alone. Furthermore, by generalizing the 1D to 2D transformation of Phoong et al., new 2D PR FBs with quincunx, hourglass, and parallelogram spectral support are obtained. These nonseparable FBs can be cascaded to realize new multiplierless PR directional FB for image processing and motion analysis. Design examples are given to demonstrate the usefulness of the proposed method.

This paper proposes an efficient class of perfect reconstruction (PR) modulated filter banks (MFB) using sum-of-powers-of-two (SOPOT) coefficients. This is based on a modified factorization of the DCT-IV matrix and the lossless lattice structure of the prototype filter, which allows the coefficients to be represented in SOPOT form without affecting the PR condition. A genetic algorithm (GA) is then used to search for these SOPOT coefficients. Design examples show that SOPOT MFB with a good frequency characteristic can be designed with very low implementation complexity. The usefulness of the approach is demonstrated with a 16 channel design example.

In this paper, an efficient method for designing perfect reconstruction (PR) two-channel finite impulse response (FIR) filter banks with low system delay is proposed. It is based on the use of nonlinear-phase FIR function in a structure previously proposed by Phoong et al. (see IEEE Trans. Signal Processing, vol.43, p.649-64, Mar. 1995). The design problem is formulated as a complex polynomial approximation problem and is solved effectively using the Remez exchange algorithm with very low design complexity. Design examples show that filter banks with flexible stopband attenuation and system delay can be readily obtained by the proposed algorithm.

A new family of two-channel low-delay filter banks and wavelet bases using the PR structure of Phoong, Kim and Vaidyanathan (1995) with sum of powers-of-two (SOPOT) coefficients are proposed. In particular, the functions alpha(z) and beta(z) in the structure are chosen as nonlinear-phase FIR and IIR filters, and the design of such multiplier-less filter banks is performed using the genetic algorithm. The proposed design method is very simple to use, and is sufficiently general to construct low-delay filter banks with flexible lengths, delays, and regularity. Several design examples are given to demonstrate the usefulness of the proposed method.

In this paper, a new family of multiplier-less modulated filter banks, called the SOPOT MFB, is presented. The coefficients of the proposed filter banks consist of sum of powers-of-two coefficients (SOPOT), which require only simple shifts and additions for implementation. The modulation matrix and the prototype filter are derived from the fast DCT-IV algorithm of Wang (1984) and a lattice structure. The design of the SOPOT MFB is performed using the genetic algorithm (GA). An 16-channel SOPOT MFB with 34 dB stopband attenuation is given as an example, and its average number of terms per SOPOT coefficient is only 2.6.

In this paper, a new approach for designing two-channel PR FIR filter banks with low system delay is proposed. It is based on the generalization of the structure previously proposed by Phoong et al. (1995). Such structurally PR filter banks are parameterized by two functions /spl alpha/(z) and /spl beta/(z) which can be chosen as linear-phase FIR or allpass functions to construct FIR/IIR filter banks with good frequency characteristics. In this paper, the more general case of using different nonlinear-phase FIR functions for /spl beta/(z) and /spl alpha/(z) is studied. As the linear-phase requirement is relaxed, higher stopband attenuation can still be achieved at low system delay. The design of the proposed low-delay filter banks is formulated as a complex polynomial approximation problem, which can be solved by the Remez exchange algorithm or analytic formula with very low complexity. The usefulness of the proposed algorithm is demonstrated by several design examples.

In this paper, a new method for designing perfect reconstruction (PR) two-channel causal stable IIR filter banks is introduced. It is based on a structure previously proposed by Phoong et al. (1995). By using a combination of allpass and linear-phase FIR functions, the bumping problem found in the conventional structural PR filter bank is significantly suppressed. The design problem is formulated as a polynomial approximation problem and is solved effectively using the Remez exchange algorithm. Filter banks with flexible stopband attenuation and system delay can readily be obtained using the proposed algorithm.

This letter proposes a method for designing a class of M-channel, causal, stable, perfect reconstruction (PR) IIR cosine-modulated filter banks (CMFB). The proposed CMFB has the same denominator for all its polyphase components in the prototype filter. Therefore, the PR condition is considerably simplified, and it is relatively simple to satisfy the PR and the causal-stable requirements of the IIR CMFB. Design examples show that the proposed IIR CMFB has sharper cutoff, higher stopband attenuation, and passband flatness than its FIR counterparts, especially when the system delay is small.

JPEG is an international standard for still image compression. The JPEG baseline algorithm allows users to supply the custom quantization table and Huffman table to control the compression ratio and the quality of the encoded image. Methods for determining the quantization matrix are usually based on (i) rate-distortion theory and (ii) spatial masking effects of the human visual system. Wu and Gersho (1993) proposed a recursive algorithm for generating picture-adaptive quantization tables based on rate-distortion approach but the complexity of the encoding algorithm is rather high. In this paper, we propose improvements to the Wu-Gersho's algorithm and a new bit allocation algorithm. Simulation results show that our new algorithm is superior to the Wu-Gersho's algorithm in terms of speed and peak signal to noise ratio (PSNR). Moreover, by incorporating the human visual system (HVS), our proposed coder can encode images with better visual quality.

Real-time image processing usually requires enormous throughput rate and huge amount of operations. Parallel processing in form of specialized hardware or multiprocessing are therefore indispensable. This paper describes a flexible programmable image processing system using Field Programmable Gate Array (FPGA). The logic cell nature of current available FPGA is most suitable for performing real-time bit-level image processing operations using bit-level systolic concept. Here, we proposed a flexible architecture, PIPS, for the integration of these programmable hardware and digital signal processors (DSP) to handle bit-level as well as arithmetic operations found in many image processing applications. The versatility of the system is demonstrated for the implementation of a 1D median filter.>