Showing papers on "Upsampling published in 2004"

Downsampling dependent upsampling of images

Abstract: Downsampling an image results in the loss of image information that cannot be recovered with upsampling. We demonstrate that the particular combination of downsampling and upsampling methods used can significantly impact the reconstructed image quality, and then we propose a technique to identify patterns associated with different downsampling methods in order to select the appropriate upsampling mechanism. The technique has low complexity and achieves high accuracy over a wide range of images.

Multi-carrier ofdm uwb communications systems

Abstract: Multiple subbands of a transmission band are continuously occupied by an OFDM signal that would otherwise occupy only a single subband. Steps may include producing an OFDM symbol (309); transforming the OFDM symbol to produce an OFDM signal (311); upsampling the OFDM signal to produce an upsampled OFDM signal (321); applying a pseudorandom code (323) to the upsampled OFDM signal to produce a coded OFDM signal; and upconverting (317) the coded OFDM signal to produce a radio frequency signal. Alternatively, steps may include upconverting the OFDM signal to produce a radio frequency signal; wherein the radio frequency signal occupies multiple ones of the following subbands: about 3200MHz to about 3700MHz; about 3700MHz to about 4200MHz; and about 4200MHz to about 4800MHz. In another embodiment, steps may include producing a sequence of N consecutive identical OFDM symbols.

Non-redundant, linear-phase, semi-orthogonal, directional complex wavelets [image/video processing applications]

Abstract: The directionality and phase information provided by nonredundant complex wavelet transforms (NCWTs) provide significant potential benefits for image/video processing and compression applications. However, because existing NCWTs are created by downsampling filtered wavelet coefficients, the finest scale of these transforms has a resolution 4/spl times/ lower than the real input signal. In this paper, we propose a linear-phase, semi-orthogonal, directional NCWT design using a novel triband filter bank. At the finest scale, the resulting transform has a resolution 3/spl times/ lower than the real input signal. We provide a design example to demonstrate three important properties for image/video processing applications: directionality, magnitude coherency, and phase coherency.

Residue image down/up sampling method and apparatus and image encoding/decoding method and apparatus using residue sampling

Abstract: A residue image down- and/or up-sampling method and apparatus and an image encoding and/or decoding method and apparatus using the residue image down- and/or up-sampling method and apparatus are provided. The residue image downsampling method includes: generating a residue corresponding to the difference between an original image and a predicted image, for each image component of the original image formed with at least two or more image components; and downsampling the residue for each image component at a predetermined ratio. The residue image upsampling method includes: upsampling data downsampled from residue data of an original image; and restoring the original image by adding the predicted image to the upsampled residue of each component. According to the methods and apparatuses, a residue image is obtained by performing spatiotemporal prediction encoding first, and by sampling this residue image, loss of information occurring in the sampling process can be reduced. Since sampling is performed with a residue image obtained through a spatiotemporal prediction process, even when an original image that is not color transformed is directly encoded, sampling can be performed effectively. Also, the methods and apparatuses have an advantage that in addition to colors, sampling of any components can be performed effectively.

Shift variance and cyclostationarity in multirate filter banks

Abstract: Multirate filtering generally introduces time-varying phenomena into the decomposed signals. These phenomena depend on whether the signals are viewed as deterministic, or as wide-sense (WS) stationary random signals. We consider the behaviour of subband energy spectra and subband energies for deterministic signals, and of subband power spectra and subband power for WS stationary random signals. We show that aliasing in decimation causes energy spectra and energies of deterministic signals to be translation variant, but leaves WS stationary random signals WS stationary. Interpolation by upsampling and anti-imaging filtering attenuates translation variance of deterministic signals, but converts WS stationary random signals into WS cyclostationary random signals. In perfect reconstruction (PR) filter banks, the time-varying effects of the filter bank channels cancel out. Subband signal processing such as filtering or quantization, however, distorts the balance between the filter channels and lets these effects appear in the output signal. Based on our analysis, we therefore derive quantitative measures for translation variance and cyclostationarity. These may be used to compare different multirate filter banks.

System and method for handling electronic ink

Abstract: A set of source points that represent a stroke input of a user is identified. The set of source points may be refined and/or modified. The set of refined/modified source points may then be stored in memory for decoding and recreation of a stroke representation. Additionally, one or both of refining and modifying the source points may be performed through one or more upsampling processes.

Digital rate converter

Abstract: A method and apparatus of converting a data signal in a digital rate converter including upsampling the input data signal at an input sampling rate to an intermediate data signal at an intermediate sampling rate, where the intermediate data signal sample values are stored in a buffer. A plurality of buffer position values are provided from a subset of buffer positions of the buffer to an interpolator, the subset of buffer positions being dependent upon a position indicator. An output data signal is provided by the interpolator at an output sampling rate, where the value of the output data signal is dependent upon a fractional indicator provided to the interpolator. The input sampling rate is based on a first clock signal and the output sampling rate is based on a second clock signal, wherein the first and second clock signal are independent of each other.

Design of Efficient Digital Interpolation Filters for Integer Upsampling

Abstract: Digital signal interpolation systems can be implemented in a variety of ways. The most basic interpolation system for integer upsampling cascades an expander unit with an interpolation lowpass filter. More complex implementations can cascade multiple expander and low-pass filter pairs. There is also flexibility in the design of interpolation filters. This thesis explores how digital interpolation systems for integer upsampling can be efficiently implemented. Efficiency is measured in terms of the number of multiplications required for each output sample point. The following factors are studied for their effect on system efficiency: the decomposition of an interpolation system into multiple cascaded stages, the use of recursive and nonrecursive interpolation filters, and the use of linear-phase and minimum-phase interpolation filters. In this thesis interpolation systems are designed to test these factors, and their computational costs are calculated. From this data, conclusions are drawn about efficient designs of interpolation systems for integer upsampling. Thesis Supervisor: Alan V. Oppenheim Title: Ford Professor of Engineering

Low-complexity sampling rate conversion method and apparatus for audio processing

Abstract: A low-complexity sampling rate conversion (SRC) method and apparatus for the processing of digital audio signals. A first stage upsamples an input audio signal to generate an upsampled audio signal. For example, the first stage may perform 1:2 upsampling using a halfband filter. A second stage re-samples the upsampled audio signal from the first stage at a target sampling rate. For example, re-sampling may be achieved using linear interpolation.

Open Loop Tracking and Inverting GPS Radio Occultation Signals: Simulation Study

Abstract: This paper estimates accuracy of the open loop tracking of the tropospheric radio occultation signals by means of numerical simulations. Each error source, such as: downsampling and integration in receiver, thermal noise, the up-sampling in postprocessing, deviation of the mean frequency of the sampled signal from zero, the length of the recorded signal, pseudo random modulation, is studied separately. The effect of each error source on the retrieved refractivity is demonstrated by use of the radio holographic method (canonical transform). This allows understanding of which error sources are dominant. Overall, the open-loop tracking of “worst case” tropospheric radio occultation signals results in sub-N unit rms refractivity retrieval errors with 100 Hz sampling and, in most cases, with 50 Hz sampling, when the signal to noise ratio (in 1 Hz band) is ≳ 700.

A Fast Algorithm Based on FFT Used in Simulation of SAR Return Wave Signal

Abstract: The simulation of distributed targets is very important to the design of synthetic aperture radar (SAR) systems Since the end product is SAR images, the echo of SAR signals needs to be simulated and then processed form SAR images, and the parameters needs to be measured and the performance of the images should be, and then evaluated The designing parameters of SAR could be adjusted to meet the requirement of users before developing SAR systems, specially in space-borne SAR systems The simulation data is also needed during the developing and debugging of a space-borne SAR real-time processor Another purpose of simulation is to supply a great deal of simulation images to train image users of space-borne SAR before it launch Since significant computation is required to simulate a large area distributed targets, it may speed a lot of time, so it is very important to research a more efficient algorithm A fast algorithm based on FFT is presented in this paper It reviews the procedure of SAR systems return wave formed A distributed target can be divided into many more facets, and each facet can be considered as a point targets with specific Radar Cross-Section (RCS) and unlimited small size Generally speaking, the SAR raw signal is the appropriate superposition of returns from each facet Noticeable, each return from every facet is the convolution of an impulse with specific amplitude and phase and the transmitting signal of SAR So we can calculate the superposition of each impulse from each facet first And then we can get the Fourier Transform (FT) of returns from all facets by multiply the Fourier Transform of impulses and transmitting signal Finally, the raw returns wave can be generated by taking an Inverse Fourier Transform (IFT) Since SAR is a discrete and band-limited system, the impulse can be calculated by an interpolation with sinc function kernel in time-domain And the Fourier Transform can be calculated using fast Fourier Transform (FFT) Interpolating error depends on the amount of interpolated points and the impulse position between two sampling points More interpolated points, less error, and impulse position is closed to any of the two sampling points, the error is decreased The relationship between the error and the two factors is presented in detail in this paper To improve the precision of interpolations, the amount of interpolated points can be increased But this method may increase computation linearly The other method is reducing the interval of sampling points Based on the second idea, an improved algorithm is introduced It refer to upsampling and extending spectrum before Interpolating, and a low-pass filter and downsampling after Fourier Transform This algorithm can improve efficiency and precision as the cost increasing its FFT computation Finally, the algorithm is implemented in a code whose performance is described and illustrated by a number of examples

Experimental characterization for the binary two-dimensional optical storage channel

Abstract: A new and challenging concept for optical storage is being developed, in which the information written on the disc has a two-dimensional characteristic, with the aim of realizing an increase of a factor of two in data density and a factor of ten in data rate over the current 3rd generation of optical storage (Blu-Ray Disc, BD). Experimental channel characterization provides insight into the magnitude of intersymbol interference, signal levels, modulation and noise distribution from the measured signal waveform. This information is used to evaluate the quality of the mastered discs, and provide feedback in the optimization process. It also yields essential input to the actual design of the signal processing solutions. The stages involved in characterizing the channel are downsampling and model fitting using a scalar diffraction model. The information gathered from channel characterization is demonstrated with experimental measurements from discs with an initial density of 1.4/spl times/BD using the same physical read-out system (blue laser and high-NA lens).

Improved interpolation of 4∶2∶0 colour images to 4∶4∶4 format exploiting inter-component correlation

Abstract: The paper deals with the interpolation problem of colour images or video whose chrominance components are down-sampled according to 4∶2∶0 scheme. Such interpolation is traditionally done using a linear 7-tap FIR filter which introduces some blurring and colour artifacts across the edges and details. The idea discussed in the paper is to improve the interpolation through adaptive filtering. The paper shows that strong higher-order correlation is present between luminance and chrominance channels regardless of the colour representation. This correlation may be exploited to support the interpolation of low resolution chrominance using fine detail information extracted from the luminance component of higher resolution. Experimental results show that the error due to the loss of high-frequency data in downsampling and subsequent upsampling is significantly reduced by the use of the proposed technique.

Polyphase representation of multirate Volterra filters

Abstract: This paper proposes a polyphase representation for Volterra filters. To derive the new realizations the well known linear polyphase theory is extended to the nonlinear case starting with Volterra filters. Both the upsampling and downsampling case are considered. As in the linear case (FIR filters) neither the input signal nor the Volterra kernels must fulfill constraints in order to be realized in polyphase form. The computational complexity could be reduced significantly because of two reasons. On the one hand all operations are performed at the low sampling rate and on the other hand some coefficients disappear in the polyphase representation.

Method and apparatus for down/up sampling residue image and image encoding/decoding method and apparatus using the same

Abstract: PROBLEM TO BE SOLVED: To provide a method and apparatus for down/up sampling a residue image , and an image encoding/decoding method and apparatus using the same in which a loss of information in a sampling step is reduced, effective sampling can be performed even when directly encoding a original image which is not color transformed, and in addition to colors, sampling of any components can be performed effectively. SOLUTION: This downsampling method including: generating a residue corresponding to the difference between an original image and a predicted image, for each image component of the original image formed with at least two or more image components; and down sampling the residue for each image component at a predetermined ratio, or this upsampling method including: upsampling data downsampled from residue data of an original image; and restoring the original image by adding the predicted image to the upsampled residue of each component. COPYRIGHT: (C)2005,JPO&NCIPI

Novel mitigation techniques for OFDM in a Doppler spread channel

Abstract: The performance of an orthogonal frequency division multiplexing (OFDM) system degrades in the Doppler spread channel due to intercarrier interference (ICI) resulting from loss of orthogonality In this paper, multirate sampling theory is applied to a standard OFDM system to reduce ICI Using the N-point FFT receiver causes circular interference from the periodic extension spectrum of the channel spectrum The periodic extension spectra due to sampling can be shifted further away from the fundamental spectrum if the sampling rate is increased (upsampling), thus resulting in less ICI for the edge subcarriers The new detection scheme recovers the data from the outer carriers first The ICI of the inner carriers is then reduced via a sequential interference cancellation technique, as detection moves from the outer carriers to the inner carriers Using this method, the error floor is reduced Upper and lower bounds on probability of error when using BPSK and QPSK modulation schemes, with the upsampling and new detection scheme cases are presented numerically Also, simulations are done to verify the analytical results

FPGA implementation of digital upconversion using distributed arithmetic FIR filters

Abstract: Distributed arithmetic (DA) is a high speed multiplication technique used for implementation of digital filters and signal upconversions. The DA is bit serial word parallel approach where throughput rate does not depend on filter length or data size. In this work a serial DA method is employed for FPGA implementation of digital component of the TIGER transmitter. A prototype has been synthesized and mapped using Xilinx Virtex II. The design with fourteen bit 100 tap FIR filter and upsampling ratio of eight takes only 18% of the device. Performance of the DA modulator is discussed with variable filter length and precision level.

Processing gesture signals

Abstract: A method of pre-processing gesture signals comprises the step of Filtering one or more signal segments by applying an infinite impulse response (IIR) filter both in a forward and in a backward temporal direction, so as to produce a band-limited gesture signal. The method may further comprise the step of matching the forward and backward initial conditions of the IIR filters to avoid any transients. Unevenly or sparsely sampled gesture signals may be subjected to the preliminary steps of interpolating the sampled signal, resampling the interpolated signal at a relatively high resampling frequency, filtering the resampled signal, and optionally downsampling the filtered signal, so as to produce a gesture signal having a well-defined sampling rate. An additional compression step may be carried out. The method may be utilized in conjunction with handwriting recognition methods.

PSNR Comparison of DCT-domain Image Resizing Methods

Abstract: Given a video frame in terms of its 88 block-DCT coefncients, we wish to obtain a downsized or upsized version of this Dame also in terms of 88 block DCT coefficients. The DCT being a linear unitary transform is distributive over matrix multiplication. This fact has been used for downsampling video frames in the DCT domains in Dugad's, Mukherjee's, and Park's methods. The downsampling and upsampling schemes combined together preserve all the low-frequency DCT coefficients of the original image. This implies tremendous savings for coding the difference between the original frame (unsampled image) and its prediction (the upsampled image).This is desirable for many applications based on scalable encoding of video. In this paper, we extend the earlier works to various DCT sizes, when we downsample and then upsample of an image by a factor of two. Through experiment, we could improve the PSM values whenever we increase the DCT block size. However, because the complexity will be also increase, we can say there is a tradeoff. The experiment result would provide important data for developing fast algorithms of compressed-domain image/video resizing.

Periodically Time-Varying Realizations of Multirate Converters for Hardware Reduction

Abstract: Realizations of multirate FIR filters are proposed using periodically time-varying (PTV) structures. By exploiting the computational redundancy of the filtering operation in a multirate filter, it is possible to implement the filter with much less hardware. In the proposed implementations, the number of multiplyand-add units is reduced by making several coecients share one multiplier-and-add unit in a periodic fashion. Specifically, each multiply-and-add unit performs dierent coecient scalings at dierent time instants within a period. Compared to the direct form realization, the proposed realizations reduce the hardware of an interpolator and a decimator by a factor of approximately U and M, respectively, while retaining the same processing speed, where U and M are the upsampling and downsampling factors, respectively. For a rational sampling rate conversion by a factor of U/M, where U and M are relatively prime, the proposed realization oers hardware reduction by a factor of approximately UM, compared to the conventional direct form.

Image encoding and decoding method using residue sampling

Abstract: A residue image down- and/or up-sampling method and apparatus and an image encoding and/or decoding method and apparatus using the residue image down-and/or up-sampling method and apparatus are provided. The residue image downsampling method includes: generating a residue corresponding to the difference between an original image and a predicted image, for each image component of the original image formed with at least two or more image components; and downsampling the residue for each image component at a predetermined ratio. The residue image upsampling method includes: upsampling data downsampled from residue data of an original image; and restoring the original image by adding the predicted image to the upsampled residue of each component. According to the methods and apparatuses, a residue image is obtained by performing spatiotemporal prediction encoding first, and by sampling this residue image, loss of information occurring in the sampling process can be reduced. Since sampling is performed with a residue image obtained through a spatiotemporal prediction process, even when an original image that is not color transformed is directly encoded, sampling can be performed effectively. Also, the methods and apparatuses have an advantage that in addition to colors, sampling of any components can be performed effectively.

Detection of a continuous pulse with unknown arrival time under low sampling rates

Abstract: The paper investigates the detection and time of arrival estimation problem for a pulse embedded in noise under low sampling rates with variation in energy-to-noise ratio (ENR) It remains a classical problem in the detection of signals in radar applications and not much literature is available on it The paper aims at investigating this issue subject to different sampling scenarios Being a composite hypothesis testing problem, it is dealt with using the generalized likelihood ratio test (GLRT) Different approaches are used for the formation of the test setup, including noise whitening, downsampling or upsampling the input signal to improve the detection performance, but an increase in signal-to-noise ratio (SNR) remains the critical factor for detector enhancement

Sampling-sensitive multiresolution hierarchy for irregular meshes

Abstract: Previous approaches of constructing multiresolution hierarchy for irregular meshes investigated how to overcome the connectivity and topology constraints during the decomposition, but did not consider the effects of sampling information on editing and signal processing operations. We propose a sampling-sensitive downsampling strategy and design a decomposition framework that produces a hierarchy of meshes with decreasing maximum sampling rates and increasingly regular vertex support sizes. The resulting mesh hierarchy has good quality triangles and enables more stable editing. The detail vectors better approximate the frequency spectrum of the mesh, thus making signal filtering more accurate.

Beam design for a ping-pong-sampled linear receive array

Abstract: Aliasing of far-field plane-wave information is avoided using only half the conventional Nyquist sample rate when sample timing is staggered on adjacent elements of a linear antenna array. The receive version of the array is examined here. The required upsampling beamformer structure is presented, and examples illustrate the pattern-design process.

Method for shortening calculation time in removing color aliasing artifact from color digital image

Abstract: PROBLEM TO BE SOLVED: To reduce color aliasing artifact from a color digital image having color pixels SOLUTION: This method provides the color digital image, prepares a luminance (L) signal and chrominance (C) signal from the color digital image, performs downsampling of the L and C signals and separates the image into a texture area and a non-texture area having a boundary line by using the downsampled L and C signals Furthermore, the method cleans the downsampled C signal in the texture area according to the boundary line of the texture area and the downsampled C signal and cleans the downsampled C signal in the non-texture area according to the downsampled C signal The color digital image from which the color aliasing artifact is reduced is obtained by performing upsampling of the C signal of which the noise is cleaned and using the L signal and the upsampled C signal COPYRIGHT: (C)2004,JPO

Lattices modeling as support for multigrid and multirate processing of two-dimensional signals

Abstract: Recently, new image processing techniques have been researched for application in many areas, such as biomedical, telecommunications, multimedia, remote sensing and optics. Multigrid and multirate processing of two-dimensional (2D) signals can be used in order to get a basis and consistent theory. Since it involves many types of lattices due to different grids geometry, they can be used as support of a consistent multirate processing theory for image applications. In this work, the concepts of lattice theory are used to make the rectangular, quincunx and hexagonal grid types modeling. The multirate processing of 2D signals involves also downsampling and upsampling and various types of sampling matrices and their formation are analyzed and grid types determined by them are shown.