Multiscale Filtering Research Articles

A New Subject-Specific Discriminative and Multi-Scale Filter Bank Tangent Space Mapping Method for Recognition of Multiclass Motor Imagery.

Objective: Tangent Space Mapping (TSM) using the geometric structure of the covariance matrices is an effective method to recognize multiclass motor imagery (MI). Compared with the traditional CSP method, the Riemann geometric method based on TSM takes into account the nonlinear information contained in the covariance matrix, and can extract more abundant and effective features. Moreover, the method is an unsupervised operation, which can reduce the time of feature extraction. However, EEG features induced by MI mental activities of different subjects are not the same, so selection of subject-specific discriminative EEG frequency components play a vital role in the recognition of multiclass MI. In order to solve the problem, a discriminative and multi-scale filter bank tangent space mapping (DMFBTSM) algorithm is proposed in this article to design the subject-specific Filter Bank (FB) so as to effectively recognize multiclass MI tasks.Methods: On the 4-class BCI competition IV-2a dataset, first, a non-parametric method of multivariate analysis of variance (MANOVA) based on the sum of squared distances is used to select discriminative frequency bands for a subject; next, a multi-scale FB is generated according to the range of these frequency bands, and then decompose multi-channel EEG of the subject into multiple sub-bands combined with several time windows. Then TSM algorithm is used to estimate Riemannian tangent space features in each sub-band and finally a liner Support Vector Machines (SVM) is used for classification.Main Results: The analysis results show that the proposed discriminative FB enhances the multi-scale TSM algorithm, improves the classification accuracy and reduces the execution time during training and testing. On the 4-class BCI competition IV-2a dataset, the average session to session classification accuracy of nine subjects reached 77.33 ± 12.3%. When the training time and the test time are similar, the average classification accuracy is 2.56% higher than the latest TSM method based on multi-scale filter bank analysis technology. When the classification accuracy is similar, the training speed is increased by more than three times, and the test speed is increased two times more. Compared with Supervised Fisher Geodesic Minimum Distance to the Mean (Supervised FGMDRM), another new variant based on Riemann geometry classifier, the average accuracy is 3.36% higher, we also compared with the latest Deep Learning method, and the average accuracy of 10-fold cross validation improved by 2.58%.Conclusion: Research shows that the proposed DMFBTSM algorithm can improve the classification accuracy of MI tasks.Significance: Compared with the MFBTSM algorithm, the algorithm proposed in this article is expected to select frequency bands with good separability for specific subject to improve the classification accuracy of multiclass MI tasks and reduce the feature dimension to reduce training time and testing time.

Transient impulses enhancement based on adaptive multi-scale improved differential filter and its application in rotating machines fault diagnosis

Transient impulses caused by local defects are critical for the fault detection of rotating machines. However, they are extremely weak and overwhelmed in the strong noise and harmonic components, making the transient features are very difficult to be extracted. This paper proposes an adaptive multi-scale improved differential filter (AMIDIF) to enhance the identification of transient impulses for rotating machine fault diagnosis. In this scheme, firstly, the AMIDIF is performed to decompose the measured signal of rotating machine into a series of multi-scale improved differential filter (MIDIF) filtered signals. Subsequently, in view of the MIDIF filtered signals exhibit varying extents of validity in revealing fault features, a weighted reconstruction method using correlation analysis is proposed in which the weighted coefficients are counted and distributed to the corresponding MIDIF filtered signals to highlight the effective MIDIF filtered signals and weaken the invalid ones. Finally, the transient impulse components of rotating machinery are obtained by multiplying the weighted coefficients and the MIDIF filtered signals under different scales. Furthermore, the fault types of rotating machines are inferred from the fault defect frequencies in the envelope spectrum of the transient impulses. Simulation analysis and experimental studies are implemented to verify the performance of the AMIDIF compared with the state-of-the-art methods including spectral kurtosis (SK), multi-scale average combination different morphological filter (ACDIF) and multi-scale morphology gradient product operation (MGPO). The results prove that the AMIDIF has excellent performance in extracting transient features for rotating machines fault diagnosis.

Road Centerline Extraction From VHR Images Using SVM and Multi-Scale Maximum Response Filter

In this work, an integrated framework comprising of pixel-based classification, road network filtering, and multi-scale Gabor filter is proposed to address the various prevailing issues in road centerline extraction from VHR images. The proposed framework is composed of three steps; generation of the initial road map, road network filtering and road centerline extraction. In the first step, pixel-based classification using support vector machine (SVM) is performed on VHR imagery to classify it into the road and non-road classes. In the road network filtering step, to retain the road features in classified imagery, an edge-preserving guided filter is applied and to improve the veracity of road extraction, undesirable components are eliminated by shape feature analysis. Finally, the complete and accurate road centerline are obtained by combining multi-scale Gabor filter and fast parallel thinning algorithm. The maximum response from a multi-scale Gabor filter is not only used for extraction but also used for reconstructing the broken road network, which is due to occlusions and artifacts. The proposed framework is implemented on publically available VHR images dataset. The simulation outcomes reflect the dominance of the proposed framework on different quantitative evaluation parameters as compared to up-to-date methods.

Dual attention per-pixel filter network for spatially varying image deblurring

Spatially varying motion deblurring has recently witnessed substantial progress due to the development of deep neural network. However, most existing CNN-based methods involve two major shortcomings: (1) The CNN weights are space-sharing, and these methods thus ignore the properties of complex spatially variant blurs which vary from pixel to pixel in natural blurry images. (2) Stacked convolution layers with a large kernel or recurrent neural networks (RNNs) cannot capture the global contextual dependence of features, they thus cannot exploit the relationship between different blur pixels at a distance. To solve these problems, we propose a new dual attention per-pixel filter network (DAPFN). First, we develop a multiscale per-pixel filter network (MSPFN) to learn a specific deblurring mapping for different blur pixels, which predicts the per-pixel spatially adaptive convolution kernel for each blur pixel in the input blurry image of different scales and restores the clean pixel by performing channel-wise spatially adaptive convolution with the local neighborhood pixels. Second, we develop a dual attention enhanced residual network (DAERN) to capture the global contextual dependence of the blurry images, which introduces a dual attention (DA) module consisting of the spatial self-attention module (SSA) and channel self-attention module (CSA). The fusion of the two attention modules helps to further improve the deblurring performance. Third, we propose a new receptive field selection (RFS) block to learn the nonlinear characteristics of spatially variant blurs, which enables the adaptive fusing of the features with different receptive fields and effectively enhances the network nonlinear representation ability. The experimental results on GOPRO dataset indicate that the average PSNR and SSIM of the proposed method reached 31.8455 and 0.9231, respectively. The results of extensive experiments pertaining to spatially varying image deblurring demonstrate that the proposed method outperforms the state-of-the-art image deblurring methods.

Hessian-MRLoG: Hessian information and multi-scale reverse LoG filter for pulmonary nodule detection

Computer-aided detection (CADe) of pulmonary nodules is an effective approach for early detection of lung cancer. However, due to the low contrast of lung computed tomography (CT) images, the interference of blood vessels and classifications, CADe has the problems of low detection rate and high false-positive rate (FPR). To solve these problems, a novel method using Hessian information and multi-scale reverse Laplacian of Gaussian (LoG) (Hessian-MRLoG) is proposed and developed in this work. Also, since the intensity distribution of the LoG operator and the lung nodule in CT images are inconsistent, and their shapes are mismatched, a multi-scale reverse Laplacian of Gaussian (MRLoG) is constructed. In addition, in order to enhance the effectiveness of target detection, the second-order partial derivatives of MRLoG are partially adjusted by introducing an adjustment factor. On this basis, the Hessian-MRLoG model is developed, and a novel elliptic filter is designed. Ultimately, in this study, the method of Hessian-MRLoG filtering is proposed and developed for pulmonary nodule detection. To verify its effectiveness and accuracy, the proposed method was used to analyze the LUNA16 dataset. The experimental results revealed that the proposed method had an accuracy of 93.6% and produced 1.0 false positives per scan (FPs/scan), indicating that the proposed method can improve the detection rate and significantly reduce the FPR. Therefore, the proposed method has the potential for application in the detection, localization and labeling of other lesion areas.

Distinguishing type II focal cortical dysplasias from normal cortex: A novel normative modeling approach

ObjectiveFocal cortical dysplasias (FCDs) are a common cause of apparently non-lesional drug-resistant focal epilepsy. Visual detection of subtle FCDs on MRI is clinically important and often challenging. In this study, we implement a set of 3D local image filters adapted from computer vision applications to characterize the appearance of normal cortex surrounding the gray-white junction. We create a normative model to serve as the basis for a novel multivariate constrained outlier approach to automated FCD detection. MethodsStandardized MPRAGE, T2 and FLAIR MR images were obtained in 15 patients with radiologically or histologically diagnosed FCDs and 30 healthy volunteers. Multiscale 3D local image filters were computed for each MR contrast then sampled onto the gray-white junction surface. Using an iterative Gaussianization procedure, we created a normative model of cortical variability in healthy volunteers, allowing for identification of outlier regions and estimates of similarity in normal cortex and FCD lesions. We used a constrained outlier approach following local normalization to automatically detect FCD lesions based on projection onto the mean FCD feature vector. ResultsFCDs as well as some normal cortical regions such as primary sensorimotor and paralimbic regions appear as outliers. Regions such as the paralimbic regions and the anterior insula have similar features to FCDs. Our constrained outlier approach allows for automated FCD detection with 80% sensitivity and 70% specificity. SignificanceA normative model using multiscale local image filters can be used to describe the normal cortical variability. Although FCDs appear similar to some cortical regions such as the anterior insula and paralimbic cortices, they can be identified using a constrained outlier detection approach. Our method for detecting outliers and estimating similarity is generic and could be extended to identification of other types of lesions or atypical cortical areas.

Deep Scattering Network With Fractional Wavelet Transform

Deep convolutional neural networks (DCNNs) have recently emerged as a powerful tool to deliver breakthrough performances in various image analysis and processing applications. However, DCNNs lack a strong theoretical foundation and require massive amounts of training data. More recently, the deep scattering network (DSN), a variant of DCNNs, has been proposed to address these issues. DSNs inherit the hierarchical structure of DCNNs, but replace data-driven linear filters with predefined fixed multi-scale wavelet filters, which facilitate an in-depth understanding of DCNNs and also offer the state-of-the-art performance in image classification. Unfortunately, DSNs suffer from a major drawback: they are suitable for stationary image textures but not non-stationary image textures, since 2D wavelets are intrinsically linear translation-invariant filters in the Fourier transform domain. The objective of this paper is to overcome this drawback using the fractional wavelet transform (FRWT) which can be viewed as a bank of linear translation-variant multi-scale filters and thus may be well suited for non-stationary texture analysis. We first propose the fractional wavelet scattering transform (FRWST) based upon the FRWT. Then, we present a generalized structure for the DSN by cascading fractional wavelet convolutions and modulus operators. Basic properties of this generalized DSN are derived, followed by a fast implementation of the generalized DSN as well as their practical applications. The theoretical derivations are validated via computer simulations.

Insulator Surface Breakage Recognition Based on Multiscale Residual Neural Network

Insulators are one of the widely used important devices in power systems, and proposing an accurate and effective online monitoring method for insulator defects is an urgent need for intelligent and informative power systems. We propose a multiscale residual neural network (MSResNN) for insulator surface breakage recognition. The proposed multiscale convolution filter uses three different size convolution kernels for convolution filtering and performs feature maps fusion, which can enrich the spatial correlation and channel correlation of feature maps. We investigated the effects of different learning rates on the model performance and confirmed that the model performs best with learning rate of $10^{-6}$ . Compared with VGG16, Inception_v3, and Resnet50, the MSResNN model has significant advantages in detection capability and occupies less computational resources, and the detection speed can meet the industrial real-time detection requirements. Through dark–light and noise experiments, it is demonstrated that the proposed method is robust and can effectively deal with low-quality insulator images.

A state-of-charge estimation method of the power lithium-ion battery in complex conditions based on adaptive square root extended Kalman filter

The control strategy of electric vehicles mainly depends on the power battery state-of-charge estimation. One of the most important issues is the power lithium-ion battery state-of-charge (SOC) estimation. Compare with the extended Kalman filter algorithm, this paper proposed a novel adaptive square root extended Kalman filter together with the Thevenin equivalent circuit model which can solve the problem of filtering divergence caused by computer rounding errors. It uses Sage-Husa adaptive filter to update the noise variable, and performs square root decomposition on the covariance matrix to ensure its non-negative definiteness. Moreover, a multi-scale dual Kalman filter algorithm is used for joint estimation of SOC and capacity; the forgetting factor recursive least-square method is used for parameter identification. To verify the feasibility of the algorithm under complicated operating conditions, different types of dynamic working conditions are performed on the ternary lithium-ion battery. The proposed algorithm has robust and accurate SOC estimation results and can eliminate computer rounding errors to improve adaptability compared to the conventional extended Kalman filter algorithm.

Three-dimensional spatial reconstruction of coronary arteries based on fusion of intravascular optical coherence tomography and coronary angiography.

We present a three-dimensional (3D) spatial reconstruction of coronary arteries based on fusion of intravascular optical coherence tomography (IVOCT) and digital subtraction angiography (DSA). Centerline of vessel in DSA images is exacted by multi-scale filtering, adaptive segmentation, morphology thinning and Dijkstra's shortest path algorithm. We apply the cross-correction between lumen shapes of IVOCT and DSA images and match their stenosis positions to realize co-registration. By matching the location and tangent direction of the vessel centerline of DSA images and segmented lumen coordinates of IVOCT along pullback path, 3D spatial models of vessel lumen are reconstructed. Using 1121 distinct positions selected from eight vessels, the correlation coefficient between 3D IVOCT model and DSA image in measuring lumen radius is 0.94% and 97.7% of the positions fall within the limit of agreement by Bland-Altman analysis, which means that the 3D spatial reconstruction IVOCT models and DSA images have high matching level.

Multi-scale Correlation Filter Tracking Algorithm Based on Feature Fusion

For the Kernel Correlation Filter (KCF) algorithm using a single feature can not fully describe the tracking target, and is susceptible to the change of the target scale, lacking the estimation of the target scale size, an improved feature fusion and multi-scale correlation filter tracking algorithm based on KCF is proposed. Firstly, in the stage of location prediction, color features and directional gradient histogram hog features are fused by feature series, and feature dimensions are added to realize target location prediction. Then the scale filter is introduced into the predicted target position to estimate the optimal target scale as the tracking result. Experimental results show that the accuracy and success rate of the improved algorithm are significantly improved compared with other classical tracking algorithms, and it can deal with scale change, occlusion, deformation and other complex situations robustly.

Constraining neutrino masses with weak-lensing multiscale peak counts

Massive neutrinos influence the background evolution of the Universe as well as the growth of structure. Being able to model this effect and constrain the sum of their masses is one of the key challenges in modern cosmology. Weak-lensing cosmological constraints will also soon reach higher levels of precision with next-generation surveys like LSST, WFIRST and Euclid. We use the MassiveNus simulations to derive constraints on the sum of neutrino masses $M_{\nu}$, the present-day total matter density $\Omega_{\rm m}$, and the primordial power spectrum normalization $A_{\rm s}$ in a tomographic setting. We measure the lensing power spectrum as second-order statistics along with peak counts as higher-order statistics on lensing convergence maps generated from the simulations. We investigate the impact of multiscale filtering approaches on cosmological parameters by employing a starlet (wavelet) filter and a concatenation of Gaussian filters. In both cases peak counts perform better than the power spectrum on the set of parameters [$M_{\nu}$, $\Omega_{\rm m}$, $A_{\rm s}$] respectively by 63$\%$, 40$\%$ and 72$\%$ when using a starlet filter and by 70$\%$, 40$\%$ and 77$\%$ when using a multiscale Gaussian. More importantly, we show that when using a multiscale approach, joining power spectrum and peaks does not add any relevant information over considering just the peaks alone. While both multiscale filters behave similarly, we find that with the starlet filter the majority of the information in the data covariance matrix is encoded in the diagonal elements; this can be an advantage when inverting the matrix, speeding up the numerical implementation.

Enhancing Spatial Steganographic Algorithm Based on Multi-Scale Filters

Enhancing Spatial Steganographic Algorithm Based on Multi-Scale Filters

A New Adaptive Multiscale Morphological Filter and Robust RVFLN Classifier for Distributed Generation Systems during Islanding and Non-Islanding Events

A new method for islanding and non-islanding disturbances detection and classification is proposed for a multiple PV based distributed generation (DG) system utilizing adaptive multi-scale morphological filter (AMF) and random vector functional link network (RVFLN) classifier. In comparison to different signal analysis techniques, the mathematical MF, that has wide application in power signals, EEG signal analysis, image processing, pattern recognition, etc. posses the benefit of easy execution, fast processing, and minimal computations. Further it is well known that a single scale morphological filter has limited noise filtering capacity and may also filter useful signal disturbance components resulting in erroneous detection of disturbance signals in microgrid. Therefore, an adaptive multiscale combined morphological filter is presented in this paper built on the concept of multiscale overall filtering which has better denoising effect and can retain useful signals better than the traditional filter. The proposed technique is built upon the measurement of voltage signal samples and the processing of these signals through AMF has been done for feature extraction. The extracted features are then employed as inputs to an efficient, fast, and easily implementable randomized network based classifier (RVFLN) which is made robust to reject the presence noise and outliers in the signal data. The outputs exhibited from the suggested technique concludes that it is s very fast and accurate technique for the detection and classification of islanding and non-islanding events in comparison to the widely used approaches.

Automatic detection of intracranial aneurysms in 3D-DSA based on a Bayesian optimized filter

BackgroundIntracranial aneurysm is a common type of cerebrovascular disease with a risk of devastating subarachnoid hemorrhage if it is ruptured. Accurate computer-aided detection of aneurysms can help doctors improve the diagnostic accuracy, and it is very helpful in reducing the risk of subarachnoid hemorrhage. Aneurysms are detected in 2D or 3D images from different modalities. 3D images can provide more vascular information than 2D images, and it is more difficult to detect. The detection performance of 2D images is related to the angle of view; it may take several angles to determine the aneurysm. As the gold standard for the diagnosis of vascular diseases, the detection on digital subtraction angiography (DSA) has more clinical value than other modalities. In this study, we proposed an adaptive multiscale filter to detect intracranial aneurysms on 3D-DSA.MethodsAdaptive aneurysm detection consists of three parts. The first part is a filter based on Hessian matrix eigenvalues, whose parameters are automatically obtained by Bayesian optimization. The second part is aneurysm extraction based on region growth and adaptive thresholding. The third part is the iterative detection strategy for multiple aneurysms.ResultsThe proposed method was quantitatively evaluated on data sets of 145 patients. The results showed a detection precision of 94.6%, and a sensitivity of 96.4% with a false-positive rate of 6.2%. Among aneurysms smaller than 5 mm, 93.9% were found. Compared with aneurysm detection on 2D-DSA, automatic detection on 3D-DSA can effectively reduce the misdiagnosis rate and obtain more accurate detection results. Compared with other modalities detection, we also get similar or better detection performance.ConclusionsThe experimental results show that the proposed method is stable and reliable for aneurysm detection, which provides an option for doctors to accurately diagnose aneurysms.

Multi-Scale Anti-Occlusion Correlation Filters Object Tracking Method Based on Complementary Features

Aiming to tackle the problem of tracking drift easily caused by complex factors during the tracking process, this paper proposes an improved object tracking method under the framework of kernel correlation filter. To achieve discriminative information that is not sensitive to object appearance change, it combines dimensionality-reduced Histogram of Oriented Gradients features and Lab color features, which can be used to exploit the complementary characteristics robustly. Based on the idea of multi-resolution pyramid theory, a multi-scale model of the object is constructed, and the optimal scale for tracking the object is found according to the confidence maps’ response peaks of different sizes. For the case that tracking failure can easily occur when there exists inappropriate updating in the model, it detects occlusion based on whether the occlusion rate of the response peak corresponding to the best object state is less than a set threshold. At the same time, Kalman filter is used to record the motion feature information of the object before occlusion, and predict the state of the object disturbed by occlusion, which can achieve robust tracking of the object affected by occlusion influence. Experimental results show the effectiveness of the proposed method in handling various internal and external interferences under challenging environments.

A variational successive corrections approach for the sea ice concentration analysis

The sea ice concentration observation from satellite remote sensing includes the spatial multi-scale information. However, traditional data assimilation methods cannot better extract the valuable information due to the complicated variability of the sea ice concentration in the marginal ice zone. A successive corrections analysis using variational optimization method, called spatial multi-scale recursive filter (SMRF), has been designed in this paper to extract multi-scale information resolved by sea ice observations. It is a combination of successive correction methods (SCM) and minimization algorithms, in which various observational scales, from longer to shorter wavelengths, can be extracted successively. As a variational objective analysis scheme, it gains the advantage over the conventional approaches that analyze all scales resolved by observations at one time, and also, the specification of parameters is more convenient. Results of single-observation experiment demonstrate that the SMRF scheme possesses a good ability in propagating observational signals. Further, it shows a superior performance in extracting multi-scale information in a two-dimensional sea ice concentration (SIC) experiment with the real observations from Special Sensor Microwave/Imager SIC (SSMI).

EOLGD: an edge feature descriptor-based method for long-wave infrared and visible image registration

Image registration is an essential prerequisite for multisource image fusion. To further improve the registration accuracy for long-wave infrared and visible images, a feature point descriptor construction method—edge-oriented log-Gabor descriptor is proposed. Initially, grayscale edge images are obtained by a structured random forests edge detector, which is more suitable for extracting features using multiscale and multioriented log-Gabor filters. Then, edge features are used for selecting orientation-stabilized feature points from speeded up robust feature points, and descriptors are constructed with a two-stage Gaussian weighting scheme. Finally, feature point pairs are matched by computing the difference of descriptors. Experimental results show that the proposed approach achieves better performance compared with state-of-the-art methods.

Evaluation and Analysis of Cardiovascular Function in Intensive Care Unit Patients by Ultrasound Image Segmentation Based on Deep Learning

Many studies have shown that cardiovascular disease has become one of the major diseases leading to death in the world. Therefore, it is a very meaningful topic to use image segmentation technology to segment blood vessels for clinical application. In order to automatically extract the features of blood vessel images in the process of segmentation, the deep learning algorithm is combined with image segmentation technology to segment the nerve cell membrane and carotid artery images of ICU patients, and to segment the blood vessel images from a multi-dimensional perspective. The relevant data are collected to observe the effect of this model. The results show that the three-dimensional multi-scale linear filter has a good effect on carotid artery segmentation in the image segmentation of nerve cell membranes and carotid artery. When analyzing the accuracy of vascular image segmentation from network parameters and training parameters, it is found that the accuracy of the threedimensional multi-scale linear filter can reach about 85%. Therefore, it can be found that the combination of deep learning algorithm and image segmentation technology has a good segmentation effect, and the segmentation accuracy is also high. The experiment achieves the desired effect, which provides experimental basis for the clinical application of the vascular image segmentation technology.

Multi-scale dyadic filter modulation based enhancement and classification of medical images

For the last many decades, the research is towards the classification of medical images in the early phase of its detection. But, the task becomes challenging due to the absence of the color information, like in natural scene images, and low illumination. In this paper, a multi-scale spectral approach is proposed for the classification of medical images. The proposed approach uses a dyadic filter bank extended to six scales for simultaneous modulation of the frequency and amplitude signal of the medical image. The modulated signal strength is used for enhancing the contrast of the image as a preprocessing step. The 32 bin spectral histogram is used to fetch the features using different modulation components at each scale of the dyadic filter bank. The proposed method has experimented with two medical imaging databases - one is malignant Brain tumor MRI scans collected from SMS medical college Jaipur. The second database is from the TCIA data repository having three datasets of Lung-CT and Brain MRI. These datasets have experimented with SVM using a quadratic kernel function. The experimental results show that the proposed approach fetches better textural information as compared with traditional texture analysis methods. Based on the analysis of the experimentation results, it is recommended that the use of the spectral features gives better early detection of the abnormalities for medical imaging datasets.