Thresholding Step Research Articles

Loopy Residual Hashing: Filling the Quantization Gap for Image Retrieval

Hashing has been widely used in large-scale image retrieval based on approximate nearest neighbor search. Most learning-to-hashing methods adopt a two-stage algorithm to generate binary codes. First, original images are mapped into continuous visual features. Then, binary codes are generated by quantization step or separate projection. Nevertheless, these methods are sensitive to quantization operation, i.e., thresholding. To explicitly address this issue, this study proposes a novel feature quantization scheme with a loopy recurrent neural network, called loopy residual hashing, for the purpose of high accuracy in image retrieval. Instead of one-off thresholding-based feature binarization, the proposed approach performs an iterative threshold-then-approximate operation, which calculates the quantization residual after each thresholding step and then imitates another round of binarization to further approximate the coding residual. The resulting sequences of binary codes possess higher representation accuracy and extensive experiments on image retrieval demonstrate its superior discriminative capability over the prior art. In the meantime, theoretical approximation error analysis is given.

Heterogeneous Multilayer Generalized Operational Perceptron

The traditional multilayer perceptron (MLP) using a McCulloch-Pitts neuron model is inherently limited to a set of neuronal activities, i.e., linear weighted sum followed by nonlinear thresholding step. Previously, generalized operational perceptron (GOP) was proposed to extend the conventional perceptron model by defining a diverse set of neuronal activities to imitate a generalized model of biological neurons. Together with GOP, a progressive operational perceptron (POP) algorithm was proposed to optimize a predefined template of multiple homogeneous layers in a layerwise manner. In this paper, we propose an efficient algorithm to learn a compact, fully heterogeneous multilayer network that allows each individual neuron, regardless of the layer, to have distinct characteristics. Based on the complexity of the problem, the proposed algorithm operates in a progressive manner on a neuronal level, searching for a compact topology, not only in terms of depth but also width, i.e., the number of neurons in each layer. The proposed algorithm is shown to outperform other related learning methods in extensive experiments on several classification problems.

Detection of hemorrhage in retinal images using linear classifiers and iterative thresholding approaches based on firefly and particle swarm optimization algorithms

We propose a novel iterative thresholding approach based on firefly and particle swarm optimization to be used for the detection of hemorrhages, one of the signs of diabetic retinopathy disease. This approach consists of the enhancement of the image using basic preprocessing methods, the segmentation of vessels with the help of Gabor and Top-hat transformation for the removal of the vessels from the image, the determination of the number of regions with hemorrhages and pixel counts in these regions using firefly algorithm (FFA) and particle swarm optimization algorithm (PSOA)-based iterative thresholding, and the detection of hemorrhages with the help of a support vector machine (SVM) and linear regression (LR)-based classifier. In the preprocessing step, color space selection, brightness and contrast adjustment, and adaptive histogram equalization are applied to enhance retinal images, respectively. In the step of segmentation, blood vessels are detected by using Gabor and Top-hat transformations and are removed from the image to avoid confusion with hemorrhagic regions in the retinal image. In the iterative thresholding step, the number of hemorrhagic regions and pixel counts in these regions are determined by using an iterative thresholding approach that generates different thresholding values with the FFA/PSOA. In the classification step, the hemorrhagic regions and pixel counts obtained by the iterative thresholding are used as inputs in the LR/SVM-based classifier. PSOA-based iterative thresholding and the SVM classifier achieved 96.7 % sensitivity, 91.4 % specificity, and 94.1 % accuracy for hemorrhage detection. Finally, the experiments show that the correct classification rates and time performances of the PSOA-based iterative thresholding algorithm are better than those of the FFA in hemorrhage detection. In addition, the proposed approach can be used as a diagnostic decision support system for detecting hemorrhages with high success rate.

A Modified JadeR for Signal Separation under Gaussian Noise

A Modified version of Joint Approximation Diagonalization Estimation of Real Signals algorithm (JADER) is proposed to enhance efficiency and speed of Blind Signal Separation (BSS). MJADER based on the mixture's dimensions minimization step, where the cumulant matrices have been estimated using a reduced-dimension observed mixture. The approach (M-JADER) is based on a threshold step, it is easy to implement, computationally efficient and faster than standard JADER about 50% where it has less running time. The comparison done under tow types of niose(semi-white Gaussian noise and Uniform noise)

On the Convergence of the SINDy Algorithm

One way to understand time-series data is to identify the underlying dynamical system which generates it. This task can be done by selecting an appropriate model and a set of parameters which best fits the dynamics while providing the simplest representation (i.e. the smallest amount of terms). One such approach is the sparse identification of nonlinear dynamics framework [6] which uses a sparsity-promoting algorithm that iterates between a partial least-squares fit and a thresholding (sparsity-promoting) step. In this work, we provide some theoretical results on the behavior and convergence of the algorithm proposed in [6]. In particular, we prove that the algorithm approximates local minimizers of an unconstrained $\ell^0$-penalized least-squares problem. From this, we provide sufficient conditions for general convergence, rate of convergence, and conditions for one-step recovery. Examples illustrate that the rates of convergence are sharp. In addition, our results extend to other algorithms related to the algorithm in [6], and provide theoretical verification to several observed phenomena.

Threshold-Guided Design and Optimization for Harris Corner Detector Architecture

High-speed corner detection is an essential step in many real-time computer vision applications, e.g., object recognition, motion analysis, and stereo matching. Hardware implementation of corner detection algorithms, such as the Harris corner detector (HCD) has become a viable solution for meeting real-time requirements of the applications. A major challenge lies in the design of power, energy and area efficient architectures that can be deployed in tightly constrained embedded systems while still meeting real-time requirements. In this paper, we proposed a bit-width optimization strategy for designing hardware-efficient HCD that exploits the thresholding step in the algorithm to determine interest points from the corner responses. The proposed strategy relies on the threshold as a guide to truncate the bit-widths of the operators at various stages of the HCD pipeline with only marginal loss of accuracy. Synthesis results based on 65-nm CMOS technology show that the proposed strategy leads to power-delay reduction of 35.2%, and area reduction of 35.4% over the baseline implementation. In addition, through careful retiming, the proposed implementation achieves over 2.2 times increase in maximum frequency while achieving an area reduction of 35.1% and power-delay reduction of 35.7% over the baseline implementation. Finally, we performed repeatability tests to show that the optimized HCD architecture achieves comparable accuracy with the baseline implementation (average decrease of repeatability is less than 0.6%).

Local Damage Detection Method Based on Distribution Distances Applied to Time-Frequency Map of Vibration Signal

Problem of the local damage detection is a crucial task in modern condition monitoring. However, most of the classical methods rely on being tested in laboratories in the isolated test stands. In case of the real world data, there are outer sources of the noise (for instance nearby machines). Such noise can affect shape of the signal and increase the level of difficulty in finding information about the fault. Thus, one can define problem of the fault detection as a finding cyclic impulses in the signal. In this paper, the authors propose a procedure for the local damage detection based on the time-frequency decomposition and distance between distributions. Local damage in bearings/gearbox provides specific response in the vibration signal. It can be further investigated via time-frequency decomposition where one can track energy distribution change in time. Applying measure of distances between distributions to absolute value of short time Fourier transform (STFT) matrix of the vibration signal, one can find deviation of subsequent samples from one used as a referential sample. Furthermore, thresholding of the absolute value of the STFT matrix, based on the log-variance of this matrix with respect to frequency allows for the enhancement of the results. In the extension of the proposed methods, we also use normalization of the spectrogram to omit additional step of the log-variance thresholding. Knowledge, which one needs to set appropriate threshold, can set this step to be difficult in the industrial application and thus, normalization of the spectrogram is proposed. Analyzed real signals were acquired from the belt conveyor driving unit in the mining facility.

Designing Catalysts for Chirality-Selective Synthesis of Single-Walled Carbon Nanotubes: Past Success and Future Opportunity.

A major obstacle for the applications of single-walled carbon nanotubes (SWNTs) in electronic devices is their structural diversity, ending in SWNTs with diverse electrical properties. Catalytic chemical vapor deposition has shown great promise in directly synthesizing high-quality SWNTs with a high selectivity to specific chirality (n, m). During the growth process, the tube-catalyst interface plays crucial roles in regulating the SWNT nucleation thermodynamics and growth kinetics, ultimately governing the SWNT chirality distribution. Starting with the introduction of SWNT growth modes, this review seeks to extend the knowledge about chirality-selective synthesis by clarifying the energetically favored SWNT cap nucleation and the threshold step for SWNT growth, which describes how the tube-catalyst interface affects both the nucleus energy and the new carbon atom incorporation. Such understandings are subsequently applied to interpret the (n, m) specific growth achieved on a variety of templates, such as SWNT segments or predefined molecular seeds, transition metal (Fe, Co and Ni)-containing catalysts at low reaction temperatures, W-based alloy catalysts, and metal carbides at relatively high reaction temperatures. The up to date achievements on chirality-controlled synthesis of SWNTs is summarized and the remaining major challenges existing in the SWNT synthesis field are discussed.

Threshold based brain tumor image segmentation

Image processing is most vital area of research and application in field of medical-imaging. Especially it is a major component in medical science. Starting from radiology to ultrasound (sonography), MRI, etc. in lots of area image is the only source of diagnosis process. Now-a-days, different types of devices are being introduced to capture the internal body parts in medical science to carry the diagnosis process correctly. However, due to various reasons, the captured images need to be tuned digitally to gain the more information. These processes involve noise reduction, segmentations, thresholding etc. . Image segmentation is a process to segment the target area of image to identify the area more prominently. There are different process are evolved to perform the segmentation process, one of which is Image thresholding. Moreover there are different tools are also introduce to perform this step of image thresholding. The recent introduced tool PSO is being used here to segment the MRI scans to identify the brain lesions using image thresholding technique.

Unsupervised Rapid Flood Mapping Using Sentinel-1 GRD SAR Images

We present a new methodology for rapid flood mapping exploiting Sentinel-1 synthetic aperture radar data. In particular, we propose the usage of ground range detected (GRD) images, i.e., preprocessed products made available by the European Space Agency, which can be quickly treated for information extraction through simple and poorly demanding algorithms. The proposed framework is based on two processing levels providing event maps with increasing resolution. The first level exploits classic co-occurrence texture measures combined with amplitude information in a fuzzy classification system avoiding the critical step of thresholding. The second level consists of a change-detection approach applied to the full resolution GRD product. The discussion is supported by several experiments demonstrating the potentiality of the proposed methodology, which is particularly oriented toward the end-user community.

Confidence Intervals for Causal Effects with Invalid Instruments by Using Two-Stage Hard Thresholding with Voting

SummaryA major challenge in instrumental variable (IV) analysis is to find instruments that are valid, or have no direct effect on the outcome and are ignorable. Typically one is unsure whether all of the putative IVs are in fact valid. We propose a general inference procedure in the presence of invalid IVs, called two-stage hard thresholding with voting. The procedure uses two hard thresholding steps to select strong instruments and to generate candidate sets of valid IVs. Voting takes the candidate sets and uses majority and plurality rules to determine the true set of valid IVs. In low dimensions with invalid instruments, our proposal correctly selects valid IVs, consistently estimates the causal effect, produces valid confidence intervals for the causal effect and has oracle optimal width, even if the so-called 50% rule or the majority rule is violated. In high dimensions, we establish nearly identical results without oracle optimality. In simulations, our proposal outperforms traditional and recent methods in the invalid IV literature. We also apply our method to reanalyse the causal effect of education on earnings.

Nuclear Norm-Based Recursive Subspace Identification for Wind Turbine Flutter Detection

Commercial wind turbine blades are progressively becoming longer and more flexible; in order to achieve load reduction, the use of shape modifying devices is currently under research. While such modifications facilitate cost reduction, they also render the blade susceptible to the unstable aeroelastic phenomenon of flutter. To be able to detect the onset of flutter, and to modify the load control algorithm accordingly, it is desirable to perform online identification of system dynamics. In this paper, a recursive subspace identification algorithm is augmented with a nuclear norm-based cost function for the rapid identification of changes in the dominant system behavior. The time-consuming singular value thresholding step involved in the identification is replaced by a fast randomized algorithm. The method developed is used to identify the changes in the dynamics of an experimental wind turbine equipped with shape-modifying actuators, and operated under controlled conditions in a wind tunnel. The proposed identification method shows high sensitivity to changes in system dynamics, and is shown capable of stably and rapidly identifying the onset of aeroelastic flutter.

A Statistical Approach to Preprocess and Enhance C-Band SAR Images in Order to Detect Automatically Marine Oil Slicks

The aim of this paper was to propose a new methodology for preprocessing and enhancing C-band synthetic aperture radar (SAR) images for the automatic detection of marine oil slicks. The proposed methodology includes three processing levels: preprocessing, thresholding, and binary cleaning. The first level is to correct the heterogeneity of brightness in SAR images caused by the non-Lambertian reflection of the radar signal on the sea surface. This heterogeneity can be justified by: the distance from the nadir (incidence angle effect), the interaction between wind direction and radar pulse, and the wide swath mode. The second level consists of a thresholding step. The third level is to clean the binary output images from noise residues. Several preprocessing and cleaning methods have been tested and evaluated by a qualification engine that compares the automatically detected patches with a training data set of manually detected dark patches. The training data set includes oil slicks and lookalikes. As a result, the “best” preprocessing method that homogenizes the brightness of C-band SAR scenes and optimizes the automatic detection of marine oil slicks is based on an adaptation to the C-band MODel. As for the cleaning process, the tested morphological methods show that small object removal followed by a morphological closing optimizes the automatic detection of marine oil slicks.

Two-level structural sparsity regularization for identifying lattices and defects in noisy images

This paper presents a regularized regression model with a two-level structural sparsity penalty applied to locate individual atoms in a noisy scanning transmission electron microscopy image (STEM). In crystals, the locations of atoms is symmetric, condensed into a few lattice groups. Therefore, by identifying the underlying lattice in a given image, individual atoms can be accurately located. We propose to formulate the identification of the lattice groups as a sparse group selection problem. Furthermore, real atomic scale images contain defects and vacancies, so atomic identification based solely on a lattice group may result in false positives and false negatives. To minimize error, model includes an individual sparsity regularization in addition to the group sparsity for a within-group selection, which results in a regression model with a two-level sparsity regularization. We propose a modification of the group orthogonal matching pursuit (gOMP) algorithm with a thresholding step to solve the atom finding problem. The convergence and statistical analyses of the proposed algorithm are presented. The proposed algorithm is also evaluated through numerical experiments with simulated images. The applicability of the algorithm on determination of atom structures and identification of imaging distortions and atomic defects was demonstrated using three real STEM images. We believe this is an important step toward automatic phase identification and assignment with the advent of genomic databases for materials.

On developing an automatic threshold applied to feature selection ensembles

Feature selection ensemble methods are a recent approach aiming at adding diversity in sets of selected features, improving performance and obtaining more robust and stable results. However, using an ensemble introduces the need for an aggregation step to combine all the output methods that confirm the ensemble. Besides, when trying to improve computational efficiency, ranking methods that order all initial features are preferred, and so an additional thresholding step is also mandatory. In this work two different ensemble designs based on ranking methods are described. The main difference between them is the order in which the combination and thresholding steps are performed. In addition, a new automatic threshold based on the combination of three data complexity measures is proposed and compared with traditional thresholding approaches based on retaining a fixed percentage of features. The behavior of these methods was tested, according to the SVM classification accuracy, with satisfactory results, for three different scenarios: synthetic datasets and two types of real datasets (where sample size is much higher than feature size, and where feature size is much higher than sample size).

Vein Pattern Decomposition (VPD) Method: a Novel Approach to Build Practical Palm Vein Based Person Authentication System

In this paper, the authors propose a novel approach for people identification using the palm vein decomposition method. The complete system consists of Image acquisition, Noise (Hair) removal & Image enhancement, Vein pattern segmentation, Skeletonization and pattern matching. The images are acquired using NIR imaging method. RoI, noise removal, image thresholding, image binarization and image skeletonization steps are carried out on each image. Feature extraction was carried out using our novel Vein Pattern Decomposition (VPD). A database of 2640 images of 55 subjects has been created for testing. The performance evaluation is carried out in terms of FAR and FRR. VPD is a novel approach used for building this practical real time system for people authentication.

Improving DOA Estimation Algorithms Using High-Resolution Quadratic Time-Frequency Distributions

This paper addresses the problem of direction of arrival (DOA) estimation and blind source separation (BSS) for nonstationary signals in the underdetermined case. These two problems are strongly related to the mixing matrix estimation problem. To deal with the nonstationary characteristics of signals, this study uses high-resolution quadratic time-frequency distributions (TFDs) to reduce cross-terms while keeping a good resolution for the construction of spatial TFDs. The main contributions of this paper are two-fold. First, the formulation of a statistical test for the noise thresholding step improves robustness and avoids the use of empirical parameters; this test performs multisource selection of the time-frequency points where the signal of interest is present. Second, an algorithm based on image processing methods performs an auto-source selection for mixing matrix estimation. The results on simulated signals demonstrate an improvement of 10 dB in terms of normalized mean square error for BSS and 7% in terms of relative error for DOA estimation over standard methods.

An efficient iterative thresholding method for image segmentation

We proposed an efficient iterative thresholding method for multi-phase image segmentation. The algorithm is based on minimizing piecewise constant Mumford–Shah functional in which the contour length (or perimeter) is approximated by a non-local multi-phase energy. The minimization problem is solved by an iterative method. Each iteration consists of computing simple convolutions followed by a thresholding step. The algorithm is easy to implement and has the optimal complexity O(Nlog⁡N) per iteration. We also show that the iterative algorithm has the total energy decaying property. We present some numerical results to show the efficiency of our method.

Analysis of Liquid Chromatography-Mass Spectrometry Data with an Elastic Net Multivariate Curve Resolution Strategy for Sparse Spectral Recovery.

Analysis of liquid chromatography-mass spectrometry (LC-MS) data requires the differentiation between a small number of relevant chemical signals and a larger amount of noise. This is often done based, at least partially, on a threshold which assumes that low intensity m/z signals arise from the noise. This eliminates low-intensity fragments, isotopes, and adducts and may exclude relevant low-intensity compounds all together. This work describes the use of multivariate curve resolution-alternating least-squares with an additional sparse regression step using elastic net (MCR-ENALS) to distinguish relevant m/z signals without the use of a harsh thresholding step, thus allowing for discovery of low-intensity m/z signals corresponding to the analytes. This strategy is demonstrated first on a unit mass analysis of amphetamines in which relevant m/z signals are found at as low as a 0.1% intensity relative to the molecular m/z peak. The incorporation of MCR-ENALS into our previously reported data reduction strategy for analysis of high-resolution LC-MS is also demonstrated. Analysis based on only 0.3% of the original data set, while retaining low-intensity isotope peaks, was accomplished without the use of thresholding, allowing for the application of MCR-ENALS to the high-resolution LC-MS data.

An Overview of Image Analysis Algorithms for License Plate Recognition

Abstract Background and purpose: We explore the problem of License Plate Recognition (LPR) to highlight a number of algorithms that can be used in image analysis problems. In management support systems using image object recognition, the intelligence resides in the statistical algorithms that can be used in various LPR steps. We describe a number of solutions, from the initial thresholding step to localization and recognition of image elements. The objective of this paper is to present a number of probabilistic approaches in LPR steps, then combine these approaches together in one system. Most LPR approaches used deterministic models that are sensitive to many uncontrolled issues like illumination, distance of vehicles from camera, processing noise etc. The essence of our approaches resides in the statistical algorithms that can accurately localize and recognize license plate. Design/Methodology/Approach: We introduce simple and inexpensive methods to solve relatively important problems, using probabilistic approaches. In these approaches, we describe a number of statistical solutions, from the initial thresholding step to localization and recognition of image elements. In the localization step, we use frequency plate signals from the images which we analyze through the Discrete Fourier Transform. Also, a probabilistic model is adopted in the recognition of plate characters. Finally, we show how to combine results from bilingual license plates like Saudi Arabia plates. Results: The algorithms provide the effectiveness for an ever-prevalent form of vehicles, building and properties management. The result shows the advantage of using the probabilistic approached in all LPR steps. The averaged classification rates when using local dataset reached 79.13%. Conclusion: An improvement of recognition rate can be achieved when there are two source of information especially of license plates that have two independent texts.