Fast Normalized Cross-Correlation

A Kalman/map filtering (KMF)-aided fast normalized cross correlation (FNCC)-based Wi-Fi fingerprinting location sensing system is proposed in this paper. Compared with conventional neighbor selection algorithms that calculate localization results with received signal strength (RSS) mean samples, the proposed FNCC algorithm makes use of all the on-line RSS samples and reference point RSS variations to achieve higher fingerprinting accuracy. The FNCC computes efficiently while maintaining the same accuracy as the basic normalized cross correlation. Additionally, a KMF is also proposed to process fingerprinting localization results. It employs a new map matching algorithm to nonlinearize the linear location prediction process of Kalman filtering (KF) that takes advantage of spatial proximities of consecutive localization results. With a calibration model integrated into an indoor map, the map matching algorithm corrects unreasonable prediction locations of the KF according to the building interior structure. Thus, more accurate prediction locations are obtained. Using these locations, the KMF considerably improves fingerprinting algorithm performance. Experimental results demonstrate that the FNCC algorithm with reduced computational complexity outperforms other neighbor selection algorithms and the KMF effectively improves location sensing accuracy by using indoor map information and spatial proximities of consecutive localization results.

Tandem mass spectrometry (MS/MS) combined with database searching is currently the most widely used method for high-throughput peptide and protein identification. Many different algorithms, scoring criteria, and statistical models have been used to identify peptides and proteins in complex biological samples, and many studies, including our own, describe the accuracy of these identifications, using at best generic terms such as "high confidence." False positive identification rates for these criteria can vary substantially with changing organisms under study, growth conditions, sequence databases, experimental protocols, and instrumentation; therefore, study-specific methods are needed to estimate the accuracy (false positive rates) of these peptide and protein identifications. We present and evaluate methods for estimating false positive identification rates based on searches of randomized databases (reversed and reshuffled). We examine the use of separate searches of a forward then a randomized database and combined searches of a randomized database appended to a forward sequence database. Estimated error rates from randomized database searches are first compared against actual error rates from MS/MS runs of known protein standards. These methods are then applied to biological samples of the model microorganism Shewanella oneidensis strain MR-1. Based on the results obtained in this study, we recommend the use of use of combined searches of a reshuffled database appended to a forward sequence database as a means providing quantitative estimates of false positive identification rates of peptides and proteins. This will allow researchers to set criteria and thresholds to achieve a desired error rate and provide the scientific community with direct and quantifiable measures of peptide and protein identification accuracy as opposed to vague assessments such as "high confidence."

The biometric de-duplication problem examines whether an input biometric sample has a corresponding match in a reference database during the enrollment process. If the input biometric is deemed to have a match, then the individual is not enrolled as a new identity in the database in order to prevent duplicate entries; otherwise, a new identity profile is created and the individual is enrolled in the system. The goal is to insure that the biometric data of an individual is associated with a single identity or label in the database. De-duplication is necessary in applications that render services to enrolled individuals. However, little to no research has been performed to examine the errors involved in a de-duplication task, and their potential consequences. We formally introduce the types of errors that may arise in biometric de-duplication, and examine whether these errors can be modeled using traditional error measures such as the false match rate, false non-match rate, false positive identification rate, and false negative identification rate. Experimental results demonstrate that de-duplication error is impacted by the order biometric samples are tested for a duplicate and that traditional error measures are not adequate for estimating empirical de-duplication error.

Interactive effects of photospread procedures (blind administration or has knowledge of suspect identity) and presentation techniques (sequential or simultaneous photospreads) on false eyewitness identification rates were examined. Of 117 subjects (57 men and 60 women), 87 served as a witness sample and 30 as a photospread administrator sample. Witnesses viewed a 20-sec. mock robbery video prior to viewing a photospread of six photographs either simultaneously or sequentially. Half of the administrators, prior to displaying the photospreads, were made aware of the photograph of the designated suspect (single-blind condition). Consistent with previous findings, sequential presentation was associated with both lower overall false identification rates as well as lower suspect designated-misidentification rates. In each case, there was evidence that administrators' knowledge increased false identification rates more in simultaneous than in sequential presentation conditions. Implications for administrations by police investigators are discussed.

In this paper we propose a fast normalized cross correlation (NCC) algorithm for pattern matching based on combining adaptive multilevel partition with the winner update scheme. This winner update scheme is applied in conjunction with an upper bound for the cross correlation derived from Cauchy-Schwarz inequality. To apply the winner update scheme, we partition the summation of cross correlation into different levels with the partition order determined by the gradient energies of the partitioned regions in the template. Thus, this winner update scheme can be employed to skip the unnecessary calculation. Experimental results show the proposed algorithm is very efficient for image matching under different lighting conditions.

As a follow up to the recommendations of the September 1995 SETAC Pellston Workshop on Whole Effluent Toxicity (WET) on test methods and appropriate endpoints, this paper will discuss the applications and statistical properties of using a statistical criterion of minimum significant difference (MSD). We examined the upper limits of acceptable MSDs as acceptance criterion in the case of normally distributed data. The implications of this approach are examined in terms of false negative rate as well as false positive rate. Results indicated that the proposed approach has reasonable statistical properties. Reproductive data from short‐term chronic WET test with Ceriodaphnia dubia tests were used to demonstrate the applications of the proposed approach. The data were collected by the North Carolina Department of Environment, Health, and Natural Resources (Raleigh, NC, USA) as part of their National Pollutant Discharge Elimination System program.

In recent years, data security incidents caused by insider threats in distributed file systems have attracted the attention of academia and industry. The most common way to detect insider threats is based on user profiles. Through analysis, we realize that based on existing user profiles are not efficient enough, and there are many false positives when a stable user profile has not yet been formed. In this work, we propose personalized user profiles and design an insider threat detection framework, which can intelligently detect insider threats for securing distributed file systems in real-time. To generate personalized user profiles, we come up with a time window-based clustering algorithm and a weighted kernel density estimation algorithm. Compared with non-personalized user profiles, both the Recall and Precision of insider threat detection based on personalized user profiles have been improved, resulting in their harmonic mean F1 increased to 96.52%. Meanwhile, to reduce the false positives of insider threat detection, we put forward operation recommendations based on user similarity to predict new operations that users will produce in the future, which can reduce the false positive rate (FPR). The FPR is reduced to 1.54% and the false positive identification rate (FPIR) is as high as 92.62%. Furthermore, to mitigate the risks caused by inaccurate authorization for users, we present user tags based on operation content and permission. The experimental results show that our proposed framework can detect insider threats more effectively and precisely, with lower FPR and high FPIR.

Data readings from wireless sensor networks (WSNs) may be abnormal due to detection of unusual phenomena, limited battery power, sensor malfunction, or noise from the communication channel. It is thus, important to detect such data anomalies available in WSNs to determine a suitable course of action. This paper proposes an integrated data compression and anomaly detection algorithm in WSNs which can detect anomalies accurately by employing half of sensor data measurement, instead of using all the sensor data measurement. The contribution of this paper centers on data compression by using Discrete Wavelet Transform (DWT) then feeding to anomaly detection by using One-Class Support Vector Machine (OCSVM). We tested our algorithm with several synthetic and real world datasets. The results showed that the proposed algorithm outperformed previous techniques in terms of near 100% detection rate and with marginal increase in false positive rates in presence of short and noise faults.

Database characteristics can affect significantly the performance of a mammography CAD scheme. Hence adequate performance comparison among different CAD schemes is not suitable since a single scheme could present different results depending on the set of chosen cases. Images in database shou ld follow a set of quality criteria, since the imaging process up to digital file. CAD schemes can not be developed without a database used to test their efficacy, but each database with particular characteristics may in fluence on the processing scheme performance. A possible solution could be using information on the imaging equipment characteristics. This work describes a preprocessing in order to compensateŽ the image degradation during the acquisition steps, assuring a better uniformityŽ relative to the images quality. Thus, poor quality images would be restored, providing therefore some independen ce on the images source to CAD schemes and allowing to reach the better possible performance. Tests performed with mammography images sets reported a 14% increase in sensitivity for microcalcifications detection. Although this result was followed by a little increase in false positive rates, simple changes in techniques pa rameters can provide the same improvement but with a reduction of the false positive detections. Keywords: Computer-aided diagnosis in Mammography, mi crocalcifications detection, mammography images database

For an efficient analysis the estimation of the noise level in images is very important to specific estimates of each modality. Moreover, it is a fundamental step and indispensable procedure for a number of image processing approaches. Especially in magnetic resonance images (MRI) due to the Rician presented in these, where the level of noise must be evaluated. In this paper a new method to estimate the noise level in MR images T1-w is proposed and compared with a known level of the ration of signal and noise presented. The advantage of this is its easiness for utilization during image acquisition and of course the adaptability of the idea of other areas of body. The correctness of the evaluation is addressed by comparison of Atlas noise free images where the level of Rician noise was artificially added and known. The main idea is the matching of same slices after registration in order to evaluate the level of noise. For evaluation of the range of noise in an image we used the signal noise ratio - SNR and a set of MRI with increasing levels of Rician noise. However, others metrics like the normalized cross correlation - NCC or the Root Mean Squared Error (RMSE) could be used as well.

To investigate the potential value of cerebroplacental ratio (CPR) at 30-34 weeks' gestation in the prediction of adverse perinatal outcome. This was a screening study in 30 780 singleton pregnancies at 30-34 weeks' gestation. Umbilical artery (UA) and fetal middle cerebral artery (MCA) pulsatility index (PI) were measured and the values were converted to multiples of the median (MoM) after adjustment from variables in maternal characteristics and medical history that affect the measurements. CPR was calculated by dividing MCA-PI MoM by UA-PI MoM. Multivariable logistic regression analysis was used to determine if measuring CPR improved the prediction of adverse perinatal outcome provided by screening with maternal characteristics, medical history and obstetric factors. The detection rate (DR) and false-positive rate (FPR) of screening by CPR were estimated for stillbirth, Cesarean section for fetal distress, umbilical arterial cord blood pH ≤ 7.0, umbilical venous cord blood pH ≤ 7.1, 5-min Apgar score < 7 and admission to the neonatal unit (NNU) and neonatal intensive care unit (NICU). There was a significant association between CPR and birth-weight Z-score. In addition to maternal characteristics, medical history and obstetric factors, measuring CPR provided a significant contribution to the prediction of arterial cord blood pH ≤ 7.0, venous cord blood pH ≤ 7.1 and admission to NNU. The performance of CPR in screening for each adverse outcome was poor, with DR of 5-11% and a FPR of about 5%. In the small subgroup of the population delivering within 2 weeks following assessment, the DR improved to 20-50%, but with a simultaneous increase in FPR to 10-23%. The performance of CPR in routine screening for adverse perinatal outcome at 30-34 weeks' gestation is poor.

To evaluate marker distribution of free beta-human chorionic gonadotrophin (fbeta-hCG) and pregnancy-associated plasma protein-A (PAPP-A) in singleton pregnancies conceived by assisted reproduction techniques (ART). In vitro fertilization (IVF) (n = 203) and intracytoplasmic sperm injection (ICSI) (n = 192) cases from a database of 14 645 first-trimester combined tests (overall study group) were selected and matched to 1164 controls for gestational age at sample date and maternal age. In the IVF group and ICSI group, lnPAPP-A was lower (IVF 6.74 vs 7.08; P = 0.0001; ICSI 6.59 vs 7.07; P = 0.0001) compared with the matched controls. Lnfbeta-hCG was lower in the IVF group (3.75 vs 3.90; P = 0.005) but not significantly different in the ICSI group (3.87 vs 3.93; P = 0.27). The computed correction factors for PAPP-A and fbeta-hCG were 1.42 and 1.17 for the IVF group and 1.56 and 1.05 for the ICSI group.The false-positive rate (FPR) in the IVF and ICSI group compared with the matched controls was higher (IVF 10.3% vs 8.6% and ICSI 10.9% vs 7.5%). In the overall age-biased [maternal age significantly lower compared with all ART and control groups] study group the FPR was 6.8%. The increase in FPR in the ART groups can be explained by decreased PAPP-A values. Therefore, an adjustment in risk analysis for Down syndrome is suggested.

Rippled-spectrum signals (RSS) are used to evaluate the resolution of the spectro-temporal structures of sounds. Resolution measurements are based on tasks that discriminate between test and reference signals. Therefore, estimates of rippled-pattern resolution may depend on both the test and reference signal types. In the present study, ripple density resolution was measured using different reference signals. The test signals were RSS with several ripple phase reversals during the signal, and the reference signals were either RSS with a constant ripple phase or were non-rippled. The spectra were centered at 2 kHz and had an ERB of 1 oct and an SPL of 70 dB. A three-alternative forced-choice procedure was combined with an adaptive stimulation procedure. With rippled reference signals, the mean ripple density resolution was 8.9 ripples/oct. With non-rippled reference signals, the mean resolution was 26.1 ripples/oct. Different contributions of excitation-pattern and temporal-processing mechanisms were hypothesized for resolution estimates obtained with rippled and non-rippled reference signals: the excitation-pattern mechanism was expected to be more effective for discriminating between rippled signals that differ in their ripple-phase patterns, whereas the temporal-processing mechanism was expected to be more effective for discriminating between rippled and non-rippled signals.

In this paper, we analyze the error probability performance of maximal ratio combining (MRC), equal gain combining (EGC), and selection combining (SC) diversity schemes with coherent BPSK signaling on Rayleigh fading channels with Gaussian channel estimation errors. We first show that, with weighting errors, the conditional probability of error is not an explicit function of the signal-to-noise ratio (SNR) at the output of the diversity combiner. We then show that averaging the conditional probability of error with the density function of the SNR at the output of the combiner yields a lower bound on the exact probability of error, independent of the underlying diversity combining scheme. Later, we derive the exact probability of error for MRC, EGC, and SC diversity schemes and show that the exact probability of error with weighting errors is very similar to the case of perfect channel estimation, but with the average SNR per diversity branch, /spl gamma/_, for the case of perfect channel estimation, replaced by the effective SNR, /spl gamma/_/sub /spl rho//, due to weighting errors, which is a function of both /spl gamma/_ and /spl rho/, the magnitude of the normalized cross correlation between the actual and the estimated channel gains. Finally, we show that, as /spl rho/ /spl rarr/ 0, the average probability of error approaches 0.5, irrespective of the order of diversity and the diversity combining rule.

An audibly distinctive test signal provided by a generator (6) is a pink noise signal at a standard level with periodic but brief interruptions (optimized for the ear's memory for level and spectral differences) acting as an identifier. The test signal is applied to a recording medium or transmission channel (14), preferably via an equalizer (16) and gain setting means (181. The test signal recovered from the recording medium or transmission channel by playback or receiving means (22) is compared and matched to a reference uninterrupted pink noise signal provided at a standard level by a generator (24). The interruptions are used to switch between the test and reference signals and to provide via equalizer (34) and speaker (28) an audible cue as to the identity of the signal heard at any moment so that a rapid confirmation and adjustment, if necessary, of recording/reproduction or transmis- sion/reception calibration can be achieved primarily by ear and, if desired, also by applying the signals to measuring instruments, such as a meter (30) and spectrum analyser (32).