Contour Estimation Research Articles

Least-mean-square estimation of new equal-loudness level contours from recent data based on a loudness perception model

Since probable large errors were suggested in 1985 in the equal-loudness level contours by Robinson and Dadson [Br. J. Appl. Phys. 7, 166–181 (1956)], which were standardized as ISO 226, a considerable amount of data on the equal-loudness relation has been accumulated. Most of the data consistently show a large discrepancy up to more than 20 phons from the contours, especially below 1 kHz. To obtain reliable contours from these new data sporadically given for some specific frequencies and phons, a model function representing the equal-loudness relation was derived from a loudness function modified by the two-stage loudness perception model. Values of the parameters of the model function were obtained by fitting the function to the experimental data. Equal-loudness level contours could be drawn by use of the model function with the parameter values interpolated along the frequency axis. The resultant equal-loudness level contours showed clear differences from those by Robinson and Dadson for all loudness levels over the whole frequency range, particularly in the frequency range lower than 1 kHz. In contrast, the contours rather resemble those given by Fletcher and Munson in 1933 and by Churcher and King in 1937 in the midfrequency range at relatively low loudness levels. [Work supported by NEDO.]

Automated facial expression recognition system using neural networks

This paper proposes an automated facial expression recognition system using neural network classifiers. First, we use the rough contour estimation routine, mathematical morphology, and point contour detection method to extract the precise contours of the eyebrows, eyes, and mouth of a face image. Then we define 30 facial characteristic points to describe the position and shape of these three facial features. Facial expressions can be described by combining different action units, which are specified by the basic muscle movements of a human face. We choose six main action units, composed of facial characteristic point movements, as the input vectors of two different neural network‐based expression classifiers including a radial basis function network and a multilayer perceptron network. Using these two networks, we have obtained recognition rates as high as 92.1% in categorizing the facial expressions neutral, anger, or happiness. Simulation results by the computer demonstrate that computers are capable of extracting high‐level or abstract information like humans

Recognition of noisy speech by a nonstationary AR HMM with gain adaptation under unknown noise

In this paper, a gain-adapted speech recognition method in unknown noise is developed in the time domain. Noise is assumed to be colored. To cope with the notable nonstationary nature of speech signals such as fricative, glides, liquids, and transition region between phones, the nonstationary autoregressive (NAR) hidden Markov model (HMM) is used for clean speech. The nonstationary AR process is represented by using polynomial functions with a linear combination of M known basis functions. When only noisy signals are available, the estimation problem of unknown noise inevitably arises. By using multiple Kalman filters, the estimation of noise model and gain contour of speech is performed.

On Local Moments

We introduce the concept of local moments for a distribution in Rp, p⩾1, at a point z∈Rp. Local moments are defined as normalized limits of the ordinary moments of a truncated version of the distribution, ignoring the probability mass falling outside a window centered at z. The limit is obtained as the size of the window converges to 0. Corresponding local sample moments are obtained via properly normalized ordinary sample moments calculated from those data falling into a small window. The most prominent local sample moments are the local sample mean which is simply the standardized mean vector of the data falling into the window, and the local covariance, which is a standardized version of the covariance matrix of the data in the window. We establish consistency with rates of convergence and asymptotic distributions for local sample moments as estimates of the local moments. First and second order local moments are of particular interest and some applications are outlined. These include locally based iterative estimation of modes and contours and the estimation of the strength of local association.

Unsupervised contour representation and estimation using B-splines and a minimum description length criterion

This paper describes a new approach to adaptive estimation of parametric deformable contours based on B-spline representations. The problem is formulated in a statistical framework with the likelihood function being derived from a region-based image model. The parameters of the image model, the contour parameters, and the B-spline parameterization order (i.e., the number of control points) are all considered unknown. The parameterization order is estimated via a minimum description length (MDL) type criterion. A deterministic iterative algorithm is developed to implement the derived contour estimation criterion, the result is an unsupervised parametric deformable contour: it adapts its degree of smoothness/complexity (number of control points) and it also estimates the observation (image) model parameters. The experiments reported in the paper, performed on synthetic and real (medical) images, confirm the adequate and good performance of the approach.

An efficient and provably correct algorithm for the multiscale estimation of image contours by means of polygonal lines

A large portion of image contours is characterized by local properties such as sharp variations of the image intensity across the contour. The integration of local image descriptors estimated by using these local properties into curvilinear descriptors is a difficult problem from a theoretical viewpoint because of the combinatorially large number of possible curvilinear descriptors. To deal with this difficulty, the notion of compressible graphs is introduced and a contour data model is defined leading to an efficient linear-time algorithm which provably recovers contours with an upper bound on the approximation error.

Interpolation of pitch contour using temporal decomposition

A new method for predicting pitch contour of a speech signal using a small number of pitch values is addressed, for the application of very low rate speech coding, relying on the correlation between phonetic evolution and pitch variations during voiced speech segments. To track the phonetic evolution and specify perceptually significant time points, Temporal Decomposition (TD) is used. TD provides information required for both determination of critical pitch values and estimation of pitch contour by detecting event functions, as interpolation paths, and their centroids, as the most steady points, in the spectral parameters space. It is shown that the proposed method reduces the amount of pitch information to about one-tenth of that in conventional frame-by-frame based techniques with less than 5% error in pitch approximation.

Concentration and Temperature Profiles of Diffusion Flames in Droptower Experiments

The paper presents measured profiles of the temperature and the oxygen concentration of a hydrogen diffusion flame and the estimation of the flame contour under gravity and microgravity

Extracting endocardial borders from sequential echocardiographic images.

The authors discuss using mathematical morphology and temporal information (TI) to improve contour accuracy in the segmentation of 2D echo images. When the input consists of a sequence of images with poor quality, TI is used to assist the contour-extraction process. Spatial information of the contour in Frame i-1 is combined with the initial contour in Frame i to give a better estimation of the initial contour in Frame i. The results indicate that adding TI improves the accuracy of the generated contour, especially when the image quality is poor. The additional process does not affect the efficiency of the algorithm.

One-pixel-wide closed boundary identification

An appropriate space of one-pixel-wide closed (OPWC) boundary configurations is explicitly defined and an automatic algorithm to obtain OPWC contour estimates from a segmented image is presented. The motivation is to obtain a reasonable starting estimate for a Markov chain Monte Carlo-based (McMC-based) boundary optimization algorithm.

Wall position and thickness estimation from sequences of echocardiographic images

Presents a new method for endocardial (inner) and epicardial (outer) contour estimation from sequences of echocardiographic images. The framework herein introduced is fine-tuned for parasternal short axis views at the papillary muscle level. The underlying model is probabilistic; it captures the relevant features of the image generation physical mechanisms and of the heart morphology. Contour sequences are assumed to be two-dimensional noncausal first-order Markov random processes; each variable has a spatial index and a temporal index. The image pixels are modeled as Rayleigh distributed random variables with means depending on their positions (inside endocardium, between endocardium and pericardium, or outside pericardium). The complete probabilistic model is built under the Bayesian framework. As estimation criterion the maximum a posteriori (MAP) is adopted. To solve the optimization problem, one is led to (joint estimation of contours and distributions' parameters), the authors introduce an algorithm herein named iterative multigrid dynamic programming (IMDP). It is a fully data-driven scheme with no ad-hoc parameters. The method is implemented on an ordinary workstation, leading to computation times compatible with operational use. Experiments with simulated and real images are presented.

Generalization of Bombieri's theorem and its applications

By using the so-called Hooley-Huxley Contour and zero density estimates for Dirichlet L -function, Bombieri's theorem is established for a class of arithmetic functions whose generating functions satisfy certain analytic conditions. As applications of our theorem, the mean value estimates of L -functions and the distribution of integers representable as sums of two squares are discussed.

A nonsmoothing approach to the estimation of vessel contours in angiograms

Accurate and fully automatic assessment of vessel (stenoses) dimensions in angiographic images has been sought as a diagnostic tool, in particular for coronary heart disease. Here, the authors propose a new technique to estimate vessel borders in angiographic images, a necessary first step of any automatic analysis system. Unlike in previous approaches, the obtained edge estimates are not artificially smoothed; this is extremely important since quantitative analysis is the goal. Another important feature of the proposed technique is that no constant background is assumed, this making it well suited for nonsubtracted angiograms. The key aspect of the authors' approach is that continuity/smoothness constraints are not used to modify the estimates directly derived from the image (which would introduce distortion) but rather to elect (without modifying) candidate estimates. Robustness against unknown background is provided by the use a morphological edge operator, instead of some linear operator (such as a matched filter) which has to assume known background and known vessel shape.

Geostatistics for Mapping Weeds, with a Canada Thistle (Cirsium arvense) Patch as a Case Study

Geostatistical methods were used to describe and map nonrandom distribution and variation (standard deviation) of shoot density and root growth across a well-established patch of Canada thistle, a perennial weed. Semivariogram functions and kriging, an interpolation method, were used to prepare isoarithmic contour maps and associated error maps. Maps consisted of interpolated contours of uniform weed density and other measured or calculated regionalized variables between measured X-Y control points, as well as maps of error (standard deviation) associated with contour estimation. Mapped regions of greatest shoot density across a patch not only had the greatest underlying root biomass and, often, greatest density of adventitious root buds, but also had more deeply growing root biomass.

Speechreading enhancement using a sinusoidal substitute for voice fundamental frequency

This study compared the effectiveness of three acoustic supplements to speechreading: the low-pass-filtered output of an electroglottograph, a variable-frequency sinusoidal substitute for voice fundamental frequency (FO), and a constant-frequency sinusoidal substitute that served as a representation of voicing. Both sinusoidal signals were synthesized at constant amplitude during periods of voicing. The sinusoidal signals were prepared off-line by a combination of automatic and manual estimation of the FO contours of video-recorded sentences. These signals were then resynchronized with the audio portions of the original recording. In 12 normally-hearing adults, the electroglottograph signal and the variable-frequency sinusoidal FO substitute both increased the number of words recognized in sentences of known topic by between 30 and 35 percentage points. The magnitude of this effect was greater for longer sentences but independent of basic speechreading ability. The constant-frequency substitute provided a 13 percentage point increase, suggesting that approximately one-third of the FO speechreading enhancement effect could be accounted for by voicing detection alone.

Gain-adapted hidden Markov models for recognition of clean and noisy speech

In applying hidden Markov modeling for recognition of speech signals, the matching of the energy contour of the signal to the energy contour of the model for that signal is normally achieved by appropriate normalization of each vector of the signal prior to both training and recognition. This approach, however, is not applicable when only noisy signals are available for recognition. A unified approach is developed for gain adaptation in recognition of clean and noisy signals. In this approach, hidden Markov models (HMMs) for gain-normalized clean signals are designed using maximum-likelihood (ML) estimates of the gain contours of the clean training sequences. The models are combined with ML estimates of the gain contours of the clean test signals, obtained from the given clean or noisy signals, in performing recognition using the maximum a posteriori decision rule. The gain-adapted training and recognition algorithms are developed for HMMs with Gaussian subsources using the expectation-minimization (EM) approach. >

The excess-mass ellipsoid

The excess-mass ellipsoid is the ellipsoid that maximizes the difference between its probability content and a constant multiple of its volume, over all ellipsoids. When an empirical distribution determines the probability content, the sample excess-mass ellipsoid is a random set that can be used in contour estimation and tests for multimodality. Algorithms for computing the ellipsoid are provided, as well as comparative simulations. The asymptotic distribution of the parameters for the sample excess-mass ellipsoid are derived. It is found that a n 1 3 normalization of the center of the ellipsoid and lengths of its axes converge in distribution to the maximizer of a Gaussian process with quadratic drift. The generalization of ellipsoids to convex sets is discussed.

Formant contour estimation in noise using a spectral estimation “confidence measure.”

Estimation of formant contours from continuous speech is a difficult problem that has required either elaborate heuristic rules or complex hidden Markov models. Under high noise levels, these methods deteriorate significantly because of the degraded performance of the spectral estimator on which they are based. This paper proposes a method of formant contour estimation using a measure of confidence for the spectral peak estimation. At high noise levels, most spectral estimation techniques result in large variance and increased spurious/missing peaks; however, it is found that a broad distribution of spectral energy can be determined more reliably using filter-bank analysis. This spectral energy distribution is utilized to differentiate regions of high and low signal-to-noise ratio in the spectrum (for white noise corruption), which in turn is used to determine a “confidence measure” for the spectral peak estimation. The addition of this confidence information to the noisy spectral peak data of a speech utterance results in clear segments of the formant contour standing out from the remaining noisy peak data. Using these segments as anchor regions, an algorithm is developed to determine the complete contours, within the constraints of the known speech properties. [Work supported by NSF.]

Synthesis maps of ultraviolet observations of neutral interstellar gas

Copernicus and IUE measurements of neutral hydrogen column densities have been taken for about 140 stars distributed between 10 and 3000 pc from the sun and plotted in several maps to directly yield estimates of contours of neutral hydrogen column density in the 5-500 x 10 to the 17th/sq cm range. These maps show that low column density neutral hydrogen is asymmetrically distributed around the sun with a 'hole' in the neutral hydrogen located in the third quadrant. This distribution mirrors that of nearby B stars associated with the Gould belt. The maps clearly show the directions in which observations at wavelengths less than 912 A will be most successful.

Correlation inequalities and contour estimates

We give a simple estimate on the probability of contours in classical ferromagnetic spin systems, based on Griffiths' or Ginibre's correlation inequalities. This includes quite general one- and two-component spin models. Some extension also holds for alln-component anisotropic or isotropic rotators.