New developments in data processing of noisy images

In some of the methods used in metallography, the signal-noise ratio is so low that direct interpretation is not possible. With regard to microanalysis, the segregation in steels is currently being studied using quantitative X-ray mapping. The acquisition time and counting statistics are limiting factors, particularly for elements in low concentration that present low count rates compared with the background level. Noise from the useful information can be filtered out by geostatistical techniques. A new commercial software, IMAGEO, has been developed by the Ecole des Mines de Paris for noise filtering and signal deconvolution. The main features of this method and the software are described with detailed application examples. With this technique, it is now possible not only to get information from images that would be unusable, but, in the case of microanalysis, to save time for elements in higher concentration by using less counting time.     

Digital unit for statistical processing of measuring information represented as a voltage

Conclusion A digital device has been developed for statistical processing of information in the form of voltages, where measurement and conversion of the information are combined with the execution of the necessary arithmetic and logical operations, whereupon the need to store the information is eliminated. The possibilities provided, i.e., execution of the v/c, m_X, D_X, w(), r _xy ^* , modes, are realized by a sufficiently simple control, which makes the device simple in construction, reliable, and convenient for servicing.Translated from Izmeritel'naya Tekhnika, No. 8, pp. 32–34, August, 1972.     

Adaptive deblurring of noisy images

We propose a practical sensor deblurring filtering method for images that are contaminated with noise. A sensor blurring function is usually modeled via a Gaussian-like function having a bell shape. The straightforward inverse function results in the magnification of noise at high frequencies. To address this issue, we apply a special spectral window to the inverse blurring function. This special window is called the power window, which is a Fourier-based smoothing window that preserves most of the spatial frequency components in the passband and attenuates quickly at the transition band. The power window is differentiable at the transition point, which gives a desired smooth property and limits the ripple effect. Utilizing the properties of the power window, we design the deblurring filter adaptively by estimating the energy of the signal and the noise of the image to determine the passband and the transition band of the filter. The deblurring filter design criteria are (a) the filter magnitude is less than 1 at the frequencies where the noise is stronger than the desired signal (the transition band), and (b) the filter magnitude is greater than 1 at the other frequencies (the passband). Therefore the adaptively designed deblurring filter is able to deblur the image by a desired amount based on the estimated or known blurring function while suppressing the noise in the output image. The deblurring filter performance is demonstrated by a human perception experiment in which 10 observers are to identify 12 military targets with 12 aspect angles. The results of comparing target identification probabilities with blurred and deblurred images and adding two levels of noise to blurred and deblurred noisy images are reported.     

Phase-unwrapping algorithm for noisy phase-map processing

Automated fringe-pattern processing is important in a great number of industrial applications, such as optical data testing and quality control. One of the main problems that arises with these processes is the automated phase unwrapping of the phase map associated with the fringe pattern. Usually the phase map presents problems such as noise, and low-modulation areas. A new phase-unwrapping algorithm with high noise immunity is presented. The algorithm is easily implemented and can process arbitrary shapes. The main features of this algorithm are the use of a queue for the processing of arbitrary shapes and a selection criterion that determines which pixels are going to be processed.     

Wavelet filtration of noisy images

Methods of noisy image filtration using wavelet transforms with real and complex basis sets have been compared. It is shown that the use of a complex wavelet transform provides more effective filtration and admits automatic optimization of the filter parameters. Optimized choice of the threshold level during filtration based on a complex wavelet transform significantly decreases the error of image reconstruction as compared to that achieved with a standard method of discrete wavelet transform employing basis sets of the Daubechies wavelet family.

     

Improved classification robustness for noisy cell images represented as principal-component projections in a hybrid recognition system

Different types of cells are recognized from their noisy images by use of a hybrid recognition system that consists of a learning principal-component analyzer and an image-classifier network. The inputs to the feed-forward backpropagation classifier are the first 15 principal components of the 10 x 10 pixel image to be classified. The classifier was trained with clear images of cells in metaphase, unburst cells, and other erroneous patterns. Experimental results show that the recognition system is robust to image scaling and rotation, as well as to image noise. Cell recognition is demonstrated for images that are corrupted with additive Gaussian noise, impulse noise, and quantization errors. We compare the performance of the hybrid recognition system with that of a conventional three-layer feed-forward backpropagation network that uses the raw image directly as input.     

Adaptive filter solution for processing lidar returns: optical parameter estimation

Joint estimation of extinction and backscatter simulated profiles from elastic-backscatter lidar return signals is tackled by means of an extended Kalman filter (EKF). First, we introduced the issue from a theoretical point of view by using both an EKF formulation and an appropriate atmospheric stochastic model; second, it is tested through extensive simulation and under simplified conditions; and, finally, a first real application is discussed. An atmospheric model including both temporal and spatial correlation features is introduced to describe approximate fluctuation statistics in the sought-after atmospheric optical parameters and hence to include a priori information in the algorithm. Provided that reasonable models are given for the filter, inversion errors are shown to depend strongly on the atmospheric condition (i.e., the visibility) and the signal-to-noise ratio along the exploration path in spite of modeling errors in the assumed statistical properties of the atmospheric optical parameters. This is of advantage in the performance of the Kalman filter because they are often the point of most concern in identification problems. In light of the adaptive behavior of the filter and the inversion results, the EKF approach promises a successful alternative to present-day nonmemory algorithms based on exponential-curve fitting or differential equation formulations such as Klett's method.     

光学图像信息处理技术在航空侦察摄影图像处理中的应用

本文论述了对航空侦察摄影图像进行光学图像信息处理的目的及方法。     

Simulations and measurements of optical images of insonified ultrasound contrast microbubbles

Ultrasound contrast agents (UCAs) are used in a clinical setting to enhance the backscattered signal from the blood pool to estimate perfusion and blood flow. The UCAs consist of encapsulated microbubbles, measuring 1-10 /spl mu/m in diameter. Acoustic characterization of UCAs is generally carried out from an ensemble of bubbles. The measured signal is a complicated summation of all signals from the individual microbubbles. Hence, characterization of a single bubble from acoustic measurements is complex. In this study, 583 optical observations of freely flowing, oscillating, individual microbubbles from an experimental UCA were analyzed. The excursions during ultrasound exposure were observed through a microscope. Images were recorded with a high frame rate camera operating at 3 MHz. Microbubbles on these images were measured offline, and maximal excursions were determined. A technique is described to determine the diameters of the bubbles observed. We compared the maximal excursions of microbubbles of the same initial size in an ultrasound field with a 500 kHz center frequency at acoustic amplitudes ranging from 0.06 MPa to 0.85 MPa. It was concluded that maximal excursions of identical bubbles can differ by 150% at low acoustic pressures (mechanical index or MI<0.2). At a high acoustic pressure (MI=1.2) an image sequence was recorded on which a bubble collapsed, but an apparently identical bubble survived.     

Characteristic distance as a structural parameter of materials in the theory of fracture

The work deals with an important theoretical problem. We analyze two basic concepts of brittle fracture: the concept of step-by-step consecutive failures of interatomic bonds at the crack tip and the concent of “reverse” fracture. We present physical interpretation and some methods for the experimental evaluation of the characteristic distance of the fracture process which can be regarded as one of the two most important parameters that control fracture processes in materials. Institute of Problems of Strength, Ukrainian Academy of Sciences, Kiev. Translated from Fizyko-Khimichna Mekhanika Materialiv, Vol. 32, No. 1, pp. 33–37, January–February, 1996.     

一种基于统计模型的改进谱减降噪算法

提出了基于语音和噪声的傅里叶系数服从统计模型分布的假设,将基于统计模型的信噪比更新和噪声更新的方法应用于谱减法,试图解决传统谱减法中存在的音乐噪声和语音失真的问题。将提出算法与多通道谱减法和基于对数的最小均方幅度谱估计方法进行客观评价分析。利用频率加权分段信噪比评价方法、语音质量感知评价及综合质量测量等3种指标进行去噪效果评价。结果表明,所提出的基于统计模型的降噪算法效果优于MBSS,且接近Log-MMSE。     

Hyperefficient detection of targets in noisy images

We compared human detection of visual targets in noisy images with that of a theoretically optimum matched filter. Using a small thin target with vertically aligned markers, we obtained hyperefficient detection as high as 91% as compared with the theoretical optimum, a value far exceeding the 30-50% value typically reported. When the markers were removed, detection efficiencies degraded to an average of 27%, even though subjects were aware that the target was always placed in the center of a reasonably small panel. Using a nine-alternative forced-choice experiment, we compared detection by human observers with a matched-filter computational observer on a trial-by-trial basis. With the markers present, when humans missed the correct panel, they most often chose the panel with the second-highest decision variable output from the computational observer, suggesting that the template-matching model is a good one. To model results without the markers, we included location uncertainty and additional noise sources in the template matching of the computational observer. A location uncertainty of only 1 pixel, corresponding to a retinal distance of approximately 12 microm, a dimension of the order of the size of the receptive field of photoreceptors, explained the psychometric data. With the marker present, the model suggests that hyperefficient detection is obtained by limiting target location uncertainty to <6 microm. Together these results give important new insights into human visual detection mechanisms.     

S-system parameter estimation for noisy metabolic profiles using newton-flow analysis

Biochemical systems are commonly modelled by systems of ordinary differential equations (ODEs). A particular class of such models called S-systems have recently gained popularity in biochemical system modelling. The parameters of an S-system are usually estimated from time-course profiles. However, finding these estimates is a difficult computational problem. Moreover, although several methods have been recently proposed to solve this problem for ideal profiles, relatively little progress has been reported for noisy profiles. We describe a special feature of a Newton-flow optimisation problem associated with S-system parameter estimation. This enables us to significantly reduce the search space, and also lends itself to parameter estimation for noisy data. We illustrate the applicability of our method by applying it to noisy time-course data synthetically produced from previously published 4- and 30-dimensional S-systems. In addition, we propose an extension of our method that allows the detection of network topologies for small S-systems. We introduce a new method for estimating S-system parameters from time-course profiles. We show that the performance of this method compares favorably with competing methods for ideal profiles, and that it also allows the determination of parameters for noisy profiles.     

Theoretical spectrum of noisy optical pulse trains

The intensity of an ideal optical pulse train is often modeled as an exact periodic repetition of a given pulse-shape function with constant amplitude and width. Therefore, the ideal intensity power spectrum is a pure line power spectrum. However, spontaneous-emission noise due to amplification media, electronic noise due to modulators, or even intentional modulations result in period-to-period fluctuations of the pulse amplitude, width, or arrival time. The power spectrum of this so-called noisy optical pulse train is then composed of a line spectrum added to a band spectrum. This study shows that the optical pulse train intensity is cyclostationary under usual assumptions on the fluctuations. This property allows us to derive the exact optical pulse train power spectrum. A general closed-form expression that takes into account the three noise manifestations (jitter, amplitude, and width modulations) is provided. Particular expressions are given for usual cases of interest such as the jitter and amplitude modulation model, for given fluctuation probability distributions, and pulse-shape functions.     

Low-level bayesian segmentation of piecewise-homogeneous noisy and textured images

We present a novel approach to image segmentation, differing from the known “simulated annealing” method in the following ways: the compound Bayesian decision rule and consequent maximal marginal a posteriori probability (MMAP) estimates of desired region labels in pixels; the two- or three-level piecewise-homogeneous Gibbs random field with constant control parameters as the probabilistic model of the images and region maps (in the general case such a model integrates the submodels of the region map, of the ideal intensities within each region, and of the noise distorting the ideal intensities); the stochastic relaxation with the constant control parameters of the Gibbs probability distribution only as a tool to obtain the samples of this field and estimate the unknown marginal a posteriori probabilities of the region labels by collecting in each pixel the histogram of labels for these samples; the like stochastic relaxation with directed variation of the control parameters of the Gibbs probability distribution as a tool to find maximal likelihood estimates of the unknown these parameters. Some experimental results are presented.     

Regularization methods for processing fringe-pattern images

A powerful technique for processing fringe-pattern images is based on Bayesian estimation theory with prior Markov random-field models. In this approach the solution of a processing problem is characterized as the minimizer of a cost function with terms that specify that the solution should be compatible with the available observations and terms that impose certain (prior) constraints on the solution. We show that, by the appropriate choice of these terms, one can use this approach in almost every processing step for accurate and robust interferogram demodulation and phase unwrapping.     

Wavelet processing of images for target detection

This article has as its goal the development and test and evaluation of wavelet-based algorithms for automatically detecting unknown anomalies in two-dimensional images. The general idea behind the work is that the class of wavelet transforms induces a so-called multiresolution analysis (MRA) in image space whereby the image of interest is naturally decomposed into a sequence of images of varying resolution (from coarse to fine resolution) in a computationally efficient manner. The anomaly detection can then be performed sequentially beginning at a coarse scale (low resolution) and proceeding to finer scales as needed. The wavelet representation thus effectively allows the user to zoom in on particular areas of interest and thus detect image anomalies in a very efficient manner. The article includes results from computer simulations testing the proposed approach against a standard energy-detection algorithm for the unknown anomalies embedded in additive Gaussian white noise. © 1996 John Wiley & Sons, Inc.