Noise Shaping

A noise-shaped pipelined ADC is presented in this paper. A minimal complexity ∆Σ modulator in the first two sub-ADCs and residue feedback in the latter stages lead to high-order noise shaping. This also leads to reduced sensitivity to analog imperfections in the front-end stage. Implemented in 0.18µm CMOS, the ADC achieves 12 ENOB with 64 MHz clock at 6X OSR while using only a 9-bit linear front-end multiplying-DAC (MDAC). The delta-sigma sub-ADCs dissipate 400µW extra power (out of 13.9mW total power) while significantly enhancing the the overall ADC linearity.

The amount of data has been increasing exponentially in every sector such as banking securities, healthcare, education, manufacturing, consumer-trade, transportation, and energy. Most of these data are noise, different in shapes, and outliers. In such cases, it is challenging to find the desired data clusters using conventional clustering algorithms. DBSCAN is a popular clustering algorithm which is widely used for noisy, arbitrary shape, and outlier data. However, its performance highly depends on the proper selection of cluster radius (Eps) and the minimum number of points (MinPts) that are required for forming clusters for the given dataset. In the case of real-world clustering problems, it is a difficult task to select the exact value of Eps and MinPts to perform the clustering on unknown datasets. To address these, this paper proposes a dynamic DBSCAN algorithm that calculates the suitable value for Eps and MinPts dynamically by which the clustering quality of the given problem will be increased. This paper evaluates the performance of the dynamic DBSCAN algorithm over seven challenging datasets. The experimental results confirm the effectiveness of the dynamic DBSCAN algorithm over the well-known clustering algorithms.

Perceptual audio coding has become an important key technology for many types of multimedia services these days. This paper provides a brief tutorial introduction into a number of issues as they arise in todayÕs low bitrate audio coders. After discussing the Temporal Noise Shaping technology in the first part of this paper, the second part will focus on the large number of possible choices for the quantization and coding methods for perceptual audio coding along with examples of real-world systems using these approaches.

In experiments of Si deposition onto the Si(111)7ϫ7 surface using scanning tunneling microscopy, we observe ''diffusing'' Si adsorbates detected as a noise-shaped pattern in addition to the previously reported ''stationary'' ones that remain at the same positions. The stable positions and diffusion energy barriers of Si atoms on the surface are obtained using molecular-orbital calculations. For one Si atom, the diffusion barrier is only several tenths of eV, while it exceeds 1 eV for two Si atoms. We propose some assignments: the ''stationary'' adsorbate for the two Si atoms and ''diffusing'' one for the single Si atom.

Image compression algorithms employ computationally expensive spatial convolutional transforms. The CMOS image sensor performs spatially compressing image quantization on the focal plane yielding digital output at a rate proportional to the mere information rate of the video. A bank of column-parallel first-order incremental DeltaSigma-modulated analog-to-digital converters (ADCs) performs column-wise distributed focal-plane oversampling of up to eight adjacent pixels and concurrent weighted average quantization. Number of samples per pixel and switched-capacitor sampling sequence order set the amplitude and sign of the pixel coefficient, respectively. A simple digital delay and adder loop performs spatial accumulation over up to eight adjacent ADC outputs during readout. This amounts to computing a two-dimensional block matrix transform with up to 8times8-pixel programmable kernel in parallel for all columns. Noise shaping reduces power dissipation below that of a conventional digi...

This paper demonstrates a new polar transmitter architecture, which uses the digitized envelope signal to control the drain voltage of a switching mode power amplifier (PA). It is based on a novel polar modulation using the constant envelope modulated signal. Among the various constant envelope modulators, the 16 modulator is chosen for its noise-shaping characteristic. It en- ables the use of a highly efficient switching amplifier with high linearity. For demonstration, the overall transmitter is implemented and tested with a code-division multiple-access IS-95A signal. The class-D and class-F amplifiers are designed and compared for the optimum operation. The experimental results show that the amplifier with small device size is suitable for this application because of the fast switching requirement. For the class-F amplifier, the measured power-added efficiency is 51.7% at 22.1 dBm and the overall efficiency (considering the amplified quantization noise) is 31%. The adjacent channel power ratios at 885 kHz and 1.98 MHz are lower than 44.9 and 55.6 dBc at the output power range from 10.8 to 22.1 dBm without any pre-distortion techniques. The overall efficiency is improved to 48.6% with a three-level quantized 16 modulator. The results clearly show that the highly efficient switching mode PA can be controlled efficiently using a digital signal from the 16 envelope modulation technique.

Thispreprinthasbeen reproducedfrom theauthor'sadvance manuscript,withoutediting,correctionsor consideration by the Review Board. The AES takes no responsibility for the contents. Additional preprintsmay be obtainedby sendingrequestand remittanceto theAudioEngineeringSociety,60 East 42nd St.,

Denoising of speech signals using a sliding discrete cosine transforms (DCT) is proposed. A minimum mean-square error (MMSE) estimator in the domain of a sliding DCT is derived. In order to provide speech processing in real time, a fast recursive algorithm for computing the sliding DCT is presented. The algorithm is based on a recursive relationship between three subsequent local DCT spectra. Extensive testing has shown that background noise in actual environment such as the helicopter cockpit can be made imperceptible by proper choice of suppression parameters.

This work presents the design of a new 2-2 programmable sigma delta modulator architecture, for different applications, this transformation design of the ΣΔ modulator low-pass, band-pass and high-pass or vice versa with loopbacks addition, which improved the linearity of the converter and reduced the quantization noise. In this work, the MASH structure enables the implementation of stable and high-order modulator. This makes low voltage and low power applications ideal. The simulation result for sigma delta modulator for biomedical applications exhibit a signal to noise ratio is 95 dB @ 250Hz bandwidth and a 75dB @ 200KHz ,85dB @1MHz for pass band modulator. The SNR is about 70dB for 5MHz bandwidth and for high pass application. This tool will allow a development contribution and characterize a system optimization set from the start while remaining at a high level of design that is suitable for electronic systems and models VHDL-AMS, RF, Biomedical.

This paper discusses noise in range data measured by a Konica Minolta Vivid 910 scanner. Previous papers considering denoising 3D mesh data have often used artificial data comprising Gaussian noise, which is independently distributed at each mesh point. Measurements of an accurately machined, almost planar test surface indicate that real scanner data does not have such properties. An initial characterisation of real scanner noise for this test surface shows that the errors are not quite Gaussian, and more importantly, exhibit significant short range correlation. This analysis yields a simple model for generating noise with similar characteristics. We also examine the effect of two typical mesh denoising algorithms on the real noise present in the test data. The results show that new denoising algorithms are required to effectively remove real scanner noise.

This paper discusses noise in range data measured by a Konica Minolta Vivid 910 scanner. Previous papers considering denoising 3D mesh data have often used artificial data comprising Gaussian noise, which is independently distributed at each mesh point. Measurements of an accurately machined, almost planar test surface indicate that real scanner data does not have such properties. An initial characterisation of real scanner noise for this test surface shows that the errors are not quite Gaussian, and more importantly, exhibit significant short range correlation. This analysis yields a simple model for generating noise with similar characteristics. We also examine the effect of two typical mesh denoising algorithms on the real noise present in the test data. The results show that new denoising algorithms are required to effectively remove real scanner noise.

A 1-1-1 MASH ΔΣ Time-to-Digital Converter (TDC) with 6ps time resolution is implemented in 0.13μm CMOS. It achieves an ENOB of 13b and a wide input range of 100ns. The TDC exhibits stability within a temperature range of -20 to 120°C, which credits to the use of an on-chip bandgap reference and passive RC oscillators. A novel delay-line assisted online calibration technique is introduced to mitigate the skew error caused by the large comparator delay. The chip consumes only 0.7mW from 1.2V supply, and occupies 0.08mm 2 area (core). I.

A novel multi-stage noise-shaping (MASH) delta-sigma (ΔΣ) Time-to-Digital Converter (TDC) structure is proposed for applications in continuous-time pulsed time-offlight (TOF) rangefinders for nuclear reactor remote sensing, which requires both high resolution and multi MGy gamma-dose radiation tolerance. The converter, implemented in 0.13 µm CMOS, achieves a time resolution of 5.6 ps and an ENOB of 11 bits. The TDC core consumes only 1.7 mW and occupies an area of 0.11 mm 2 . Owing to the usage of circuit level radiation hardened-by-design techniques, such as passive RC oscillators and constant-g m biasing, the TDC exhibits enhanced radiation tolerance. After a total dose of 3.4 MGy at a high dose rate of 30 kGy/h, the TDC still achieves a time resolution of 10.5 ps. In order to secure the temperature stability of the TDC, a bandgap reference is employed to provide reference current and voltage for the TDC core. The total-ionizing-dose (TID) radiation tolerance of bandgap references in deep-submicron CMOS technology is generally limited by the radiation introduced leakage current in diodes. An analysis of this phenomenon is given in this paper, and a dynamic base leakage compensation (DBLC) technique is proposed to improve the radiation hardness of a bandgap reference built in a standard 0.13 µm CMOS technology. A temperature coefficient of 15 ppm/°C from -40 °C to 125 °C is measured before irradiation. The voltage variation from 0 °C to 100 °C is only ±1 mV for an output voltage of 600 mV. Gamma irradiation assessment approves that the bandgap reference is tolerant to a total ionizing dose of at least 4.5 MGy. The output reference voltage exhibits a variation of less than 1 % during the entire experiment, when the chip is irradiated by gamma ray at a dose rate of 27 kGy/h.

In this study, we develop a new shape descriptor and a matching algorithm in order to find a given template shape in an edge detected image without extracting the boundary. The shape descriptor based on Generalized Beam Angle Statistics (GBAS) defines the angles between the lines connecting each boundary point with the rest of the points, as random variable. Then, it assigns a feature vector to each point using the moments of beam angles. The proposed matching algorithm performs shape recognition by matching the feature vectors of boundary points on the template shape and the edge pixels on the image. The matching process also considers the spatial distance of the edge pixels. The experiments performed on MPEG-7 data set show that the template shapes are found succesfully on the noisy images.

System performance of an oversampled analog to digital converter (ADC) with feedback noise shaping is limited by the precision of the digital to analog converter (DAC) in the feedback path as well as by the number of integrators in the loop(s): Standard designs avoid the DAC precision problem by restricting the ADC and DAC to a single bit while stability and matching considerations limit systems to three loops. This limit in turn defines the oversample ratio for a given effective bandwidth and noise performance. We present a simple modification to the oversampled ADC which avoids these limitations via requantization in the feedback path of the original delta sigma loop structure. This modification results in greater dynamic range than is available from standard configurations.

a set of samples which have small diflerences due to noisy pixels. The learned models can be used to identify objects in outdoors i m g e s irrespectively of their size and partial occlusions. Some results of the inference procedure are shown in the pagel:

Sigma-delta modulators use a noise-shaping technique to curtail the noise power in the band of interest during digital-to-analog conversion. Error feedback modulator employs an efficient noise transfer function for time varying inputs than any other sigma-delta modulators. However, the efficiency of the conventional noise transfer function degrades and the quantizer saturation issue provokes when the input signal reaches to full scale. This work proposes a new noise transfer function which is a combination of transfer functions of two-stage Feed-forward delays and a novel Hybrid multi-stage noise shaping-error feedback sigma-delta modulator. The noise transfer function of two-stage Feed-forward delays mitigates the concern of quantizer saturation. The noise transfer function offered by the Hybrid multi-stage noise shaping-error feedback architecture provides sustainable solutions to limit cycles and idle tones. The simulation concludes that the proposed noise-shaping approach obtain...

This paper develops the noise-counterparts to familiar delay formulas like Elmore or PRIMO. By matching the first few moments of the network' s transfer impedance, we obtain efficient and accurate predictions for maximum noise between two capacitively coupled RC networks. Unlike many crosstalk equations in the literature, the method applies to general topologies and models transition-time dependence as well. Efficient enough for large circuits, the moment-matching noise formulas developed here can serve as a key ingredient in CAD methodologies that ensure a layout is free of noise problems.

In this paper detailed study of SigmaDelta (Σ-Δ) modulator has been discussed and analyzed. Low power Σ-Δ ADC for various applications has been discussed. Use of oversampling ratio in designing of Σ-Δ ADC for reduction of noise as well as for better resolution has been analyzed theoretically and practically. Keywords— SigmaDelta (Σ-Δ) modulator, Oversampled ADC, Resolution.

We address the problem of recognition of generic objects from a single intensity image. This precludes the use of purely geometric methods which assume that models are geometrically and precisely designed. Instead, we propose to use descriptions in terms of features and their qualitative geometric relationships. We propose to detect groups using perceptual organization criteria such as pro~mlty, symmetry, parallelism, and closure. The detection of these features is performed in an efficient way using proximity indexing. Since many groups are created, we also perform selection of relevant groups by orgnnlzing them into sets of similar perceptual content. Finally we present an initial implementation of a recognition system using these sets as primitives. It is an efficient colored graph matching algorithm using the adjacency matrix representation of a graph. Using indexing, we retrieve matching hypotheses, which are verified against each other with respect to topological constraints. Groups of consistent hypotheses represent detected model instances in a scene. The compIete system is illustrated on real images. We also discuss further extensions.

We address the problem of recognition of generic objects from a single intensity image. This precludes the use of purely geometric methods which assume that models are geometrically and precisely designed. Instead, we propose to use descriptions in terms of features and their qualitative geometric relationships. To succeed, it is clear that these features need to be high level, rather than points or lines. We propose to detect groups using perceptual organization criteria such as proximity, symmetry, parallelism, and closure. The detection of these features is performed in an efficient way using proximity indexing. Since many groups are created, we also perform selection of relevant groups by organizing them into sets of similar perceptual content. Finally we present an implementation of a recognition system using these sets as primitives. It is an efficient colored graph matching algorithm using the adjacency matrix representation of a graph. Using indexing, we retrieve matching hypotheses, which are verified against each other with respect to topological constraints. Groups of consistent hypotheses represent detected model instances in a scene. The complete system is illustrated on real images. We also discuss further extensions.

We address the problem of recognition of generic objects from a single intensity image. This precludes the use of purely geometric methods which assume that models are geometrically and precisely designed. Instead, we propose to use descriptions in terms of features and their qualitative geometric relationships. To succeed, it is clear that these features need to be high level, rather than points or lines. We propose to detect groups using perceptual organization criteria such as proximity, symmetry, parallelism, and closure. The detection of these features is performed in an efficient way using proximity indexing. Since many groups are created, we also perform selection of relevant groups by organizing them into sets of similar perceptual content. Finally we present an implementation of a recognition system using these sets as primitives. It is an efficient colored graph matching algorithm using the adjacency matrix representation of a graph. Using indexing, we retrieve matching hypotheses, which are verified against each other with respect to topological constraints. Groups of consistent hypotheses represent detected model instances in a scene. The complete system is illustrated on real images. We also discuss further extensions.

Synthesizing the image of a 3‐D scene as it would be captured by a camera from an arbitrary viewpoint is a central problem in Computer Graphics. Given a complete 3‐D model, it is possible to render the scene from any viewpoint. The construction of models is a tedious task. Here, we propose to bypass the model construction phase altogether, and to generate images of a 3‐D scene from any novel viewpoint from prestored images. Unlike methods presented so far, we propose to completely avoid inferring and reasoning in 3‐D by using projective invariants. These invariants are derived from corresponding points in the prestored images. The correspondences between features are established off‐line in a semi‐automated way. It is then possible to generate wireframe animation in real time on a standard computing platform. Well understood texture mapping methods can be applied to the wireframes to realistically render new images from the prestored ones. The method proposed here should allow the i...

Many classes of images possess a strong degree of spatial stationarity such that particular features of the image reliably appear in certain regions of the image. This spatial information can be used to improve compression. Unbalanced or Pruned Tree-structured Vector Quantization (PTSVQ) is a variable-rate coding technique that tends to use more bits to code active regions of the image, and fewer to code homogeneous ones. The P T S V Q is developed based on a training sequence of typical images. We used a regression tree algorithm to segment the images of the training sequence, using the t, y pitel location as a predictor f o r the intensity. This segmentation was used t o partition the training data by region and generate separate codebooks for each region, and to allocate diflering numbers of bits to the regions. Unlike other varieties of classzfied vector quantization, a region-based classification requires no side information as the decoder kiiows where in the image the current encoded block originated. These methods can enhance the perceptual quality of compressed images when compared with ordinary PTSVQ. Results are shown on magnetic resonance data.

Recent work has shown that higher-order single-bit sigma-delta modulators suffer from low-level artifacts such as idle tones and noise modulation. Techniques that have been proposed to reduce or eliminate these errors include the application of dither inside the one-bit quantiser loop, and selecting a loop filter which makes the modulator chaotic. This paper compares the efficacy of these two approaches by simulating high-resolution sigma-delta modulators suitable for audio-conversion applications.

Performances of ∆Σ modulators are evaluated by applying a coherently sampled tone and by estimating powers of the in-band tones and noise. In particular, the power of the shaped noise is usually estimated by subtracting the evaluated input tone from the output data and by integrating the power spectral density estimated by means of the periodogram. Although the coherency, the finite number of processed samples induces spectral leaking of the wideband noise, thus affecting the noise power estimate. To cope with such an issue, usually data are weighted by the Hanning sequence. In this paper, the noise power estimation error induced by the use of such window is investigated, and a criterion for choosing the minimum number of samples N which bounds the relative leakage error within a specified maximum value is explicitly given. Moreover, it is shown that, for any N , such an error is negligible for modulator orders lower than 3. Higher order modulators require the use of a large number of samples to bound the relative error of the noise power estimate when high oversampling ratios are employed.

This paper investigates the impact of clock jitter induced by substrate noise on the performance of the oversampling 16 modulators. First, a new stochastic model for substrate noise is proposed. This model is then utilized to study the clock jitter in clock generators incorporating phase-locked loops (PLLs). Next, the effect of the clock jitter on the performance of the 16 modulator is studied. It will be shown that substrate noise degrades the signal-to-noise ratio of the 16 modulator while the noise shaping does not have any effect on clock jitter induced by substrate noise. To verify the analysis experimentally, a circuit consisting of a second-order 16 modulator, a charge-pump PLL, and forty multistage digital tapered inverters driving 1-pF capacitors is designed in a 0.25-m standard CMOS process. Several experiments on the designed circuit demonstrate the high accuracy of the proposed analytical models.

Non-photorealistic rendering (NPR) has been used to produce stylized images, e.g., in a stippled or painted style. To evaluate NPR algorithms, similarity measurements used in image processing have been employed to assess the quality of rendered images. However, there is no standard objective measurement of stylization quality. In many cases, raw side-by-side comparisons are used to demonstrate improvements in aesthetic quality. This means of comparison often fails to be persuasive due to the small size of demonstrations and the subjective choice of images. We conducted a user study and examined responses of 30 subjects in order to determine two things: whether there exists a relationship between the structural quality and aesthetic quality of non-colored non-photorealistic images; and whether the choice of images matters for side-by-side comparisons. Our study revealed a statistically significant correlation between the aesthetic and structure ratings given by participants: increases in structural rating coincided with increases in aesthetic rating. Second, participants' ratings of structure and aesthetic were influenced by image content: that is, choice of input images influenced the results of side-by-side comparisons.