Mapping Module Research Articles

Deep Learning and Laser-Based 3-D Pixel-Level Rail Surface Defect Detection Method

Rail surface defect inspection is of particular importance in modern railways. Accurate and efficient surface defect detection approaches support optimized maintenance. This enables safe operation of the railway network. However, the scale and harsh working environments of the railway still pose challenges to existing manual and vision-based inspection methods. Inspired by recent advances in laser measurement and deep learning in computer vision, this paper proposes a laser-based 3D pixel-level rail surface defect detection method that combines high-precision laser measurement data with the concept of deep semantic segmentation. In the proposed method, the rail surface is first measured in 3D using a low-cost 2D laser triangulation sensor. Then, a new deep semantic segmentation network is introduced. The network is composed of a fully convolutional segmentation module and two symmetric mapping modules, which can take 3D laser measurement data as input and output 3D pixel-level defect detection results in an end-to-end manner. The modular design of the network allows the use of various segmentation modules for different applications or scenarios. Experiments on a 3D rail dataset demonstrate the feasibility of the proposed method with a pixel-level detection accuracy measured by mean Intersection over Union (mIoU) of up to 87.9%. The 3D output provides not only location and boundary information but also the 3D characterization of defects, giving an essential reference for further defect management and repair tasks.

VLSI Implementation of Optimized 2D SIMM Chaotic Map for Image Encryption

The current research work proposed a novel optimization-based 2D-SIMM (Two-Dimensional Sine Iterative chaotic map with infinite collapse Modulation Map) model for image encryption. The proposed 2D-SIMM model is derived out of sine map and Iterative Chaotic Map with Infinite Collapse (ICMIC). In this technique, scrambling effect is achieved with the help of Chaotic Shift Transform (CST). Chaotic Shift Transform is used to change the value of pixels in the input image while the substituted value is cyclically shifted according to the chaotic sequence generated by 2D-SIMM model. These chaotic sequences, generated using 2D-SIMM model, are sensitive to initial conditions. In the proposed algorithm, these initial conditions are optimized using JAYA optimization algorithm. Correlation coefficient and entropy are considered as fitness functions in this study to evaluate the best solution for initial conditions. The simulation results clearly shows that the proposed algorithm achieved a better performance over existing algorithms. In addition, the VLSI implementation of the proposed algorithm was also carried out using Xilinx system generator. With optimization, the correlation coefficient was −0.014096 and without optimization, it was 0.002585.

Reversible attack based on adversarial perturbation and reversible data hiding in YUV colorspace

Recent research on using adversarial perturbation to prevent malicious models from accessing image data has led to the corruption of image data, making images useless in other fields, especially in digital forensics. To prevent malicious models from retrieving images and ensure that authorized models can recover original image data without distortion, the reversible attack technique is rising. However, attack ability, reversibility, and image visual quality are three major challenges for existing reversible attack techniques. In this paper, a novel reversible attack method based on adversarial perturbation and reversible data hiding in YUV colorspace is proposed. We first add adversarial perturbation into the luminance channel. Then, the luminance channel distortion caused by adversarial perturbation is embedded into chrominance channels by reversible data hiding to achieve the reversible attack. In particular, the class activation mapping module is introduced to narrow the perturbation region to reduce the amount of embedded data. Experimental results on the ImageNet dataset demonstrated that the proposed method achieves better attack ability and image visual quality and ensures that original images can be recovered without distortion.

XtoE: A Novel Constructive and Camouflaged Adaptive Data Hiding and Image Encryption Scheme for High Dynamic Range Images

High dynamic range (HDR) image data hiding and encryption has attracted much interest in recent years due the benefits of providing high quality realistic images and versatile applications, such as copyright protection, data integrity, and covert communication. In this paper, we propose a novel constructive and camouflaged adaptive data hiding and image encryption scheme for HDR images. Our algorithm disguises hidden messages when converting an original OpenEXR format to the RGBE encoding, which contains the Red, Green, and Blue color channels and an exponent E channel. During the conversion process, we determine an optimal base for each pixel by considering the user’s demands and the exponent E channel information to achieve adaptive message concealment. To prevent inappropriate access to the stego image, we perform the bit-level permutation and confusion using a 2D Sine Logistic modulation map with hyperchaotic behavior and a random permutation scheme with the time complexity of ON. To the best of our knowledge, our algorithm is the first in HDR data hiding literature able to predict the image distortion and satisfy a user’s request for the embedding capacity. Our algorithm offers 18% to 32% larger embedding rate than that provided by the current state-of-the-art works without degrading the quality of the stego image. Experimental results confirm that our scheme provides high security superior to the competitors.

Machine learning-based predictions of gamma passing rates for virtual specific-plan verification based on modulation maps, monitor unit profiles, and composite dose images

Machine learning (ML) methods have been implemented in radiotherapy to aid virtual specific-plan verification protocols, predicting gamma passing rates (GPR) based on calculated modulation complexity metrics because of their direct relation to dose deliverability. Nevertheless, these metrics might not comprehensively represent the modulation complexity, and automatically extracted features from alternative predictors associated with modulation complexity are needed. For this reason, three convolutional neural networks (CNN) based models were trained to predict GPR values (regression and classification), using respectively three predictors: (1) the modulation maps (MM) from the multi-leaf collimator, (2) the relative monitor units per control point profile (MUcp), and (3) the composite dose image (CDI) used for portal dosimetry, from 1024 anonymized prostate plans. The models’ performance was assessed for classification and regression by the area under the receiver operator characteristic curve (AUC_ROC) and Spearman’s correlation coefficient (r). Finally, four hybrid models were designed using all possible combinations of the three predictors. The prediction performance for the CNN-models using single predictors (MM, MUcp, and CDI) were AUC_ROC = 0.84 ± 0.03, 0.77 ± 0.07, 0.75 ± 0.04, and r = 0.6, 0.5, 0.7. Contrastingly, the hybrid models (MM + MUcp, MM + CDI, MUcp+CDI, MM + MUcp+CDI) performance were AUC_ROC = 0.94 ± 0.03, 0.85 ± 0.06, 0.89 ± 0.06, 0.91 ± 0.03, and r = 0.7, 0.5, 0.6, 0.7. The MP, MUcp, and CDI are suitable predictors for dose deliverability models implementing ML methods. Additionally, hybrid models are susceptible to improving their prediction performance, including two or more input predictors.

Flexible Quadrature Spatial Pulse Amplitude Modulation for VLC Systems

Quadrature-spatial modulation (QSM) offers high spectral efficiency (SE) without interchannel interference for both radio frequency and visible light communication (VLC) systems. In this article, a new QSM scheme called flexible quadrature spatial pulse amplitude modulation (FQSPAM) is proposed for VLC systems. In space modulation techniques (SMTs), the data bit stream is divided into two groups as the index bits and the signal bits. These grouped bits are mapped independently to modulation symbols and indices of light-emitting diodes (LEDs). However, this mapping strategy puts a constraint on the number of LEDs and the size of the signal constellation. FQSPAM jointly designs the signal and spatial components of the constellation to overcome these limitations. This approach removes the constraint on the number of LEDs and the size of the signal constellation. The constellation design problem is formulated as a convex optimization problem to identify power-efficient constellations with different <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$M$</tex-math></inline-formula> values. A dimming-controlled optimization algorithm has also been proposed to take into account both illumination and communication at the same time. Analytical and numerical results show the improved bit error rate performance of the proposed FQSPAM compared to spatial pulse amplitude modulation and channel adaptive bit mapping.

Lossless Image Steganography Based on Invertible Neural Networks.

Image steganography is a scheme that hides secret information in a cover image without being perceived. Most of the existing steganography methods are more concerned about the visual similarity between the stego image and the cover image, and they ignore the recovery accuracy of secret information. In this paper, the steganography method based on invertible neural networks is proposed, which can generate stego images with high invisibility and security and can achieve lossless recovery for secret information. In addition, this paper introduces a mapping module that can compress information actually embedded to improve the quality of the stego image and its antidetection ability. In order to restore message and prevent loss, the secret information is converted into a binary sequence and then embedded in the cover image through the forward operation of the invertible neural networks. This information will then be recovered from the stego image through the inverse operation of the invertible neural networks. Experimental results show that the proposed method in this paper has achieved competitive results in the visual quality and safety of stego images and achieved 100% accuracy in information extraction.

Nonparametric Background Model-Based LiDAR SLAM in Highly Dynamic Urban Environments

In urban environments, simultaneous localization and mapping (SLAM) are essential for autonomous driving. Most light detection and ranging (LiDAR) SLAM methodologies have been developed for relatively static environments, despite real-world environments having many dynamic objects such as vehicles, bicycles, and pedestrians. This paper proposes an efficient and robust LiDAR SLAM. Our SLAM framework leverages the estimated background model to achieve robust motion estimation in dynamic urban environments. Based on probabilistic object estimation, the dynamic removal module estimates a nonparametric background model to recognize dynamic objects. This module estimates the probability of the difference of the range values from the accumulated LiDAR frames. Then, dynamic objects are removed by adapting the sensor velocity from the estimated ego-motion. In the local mapping module, our method optimizes the LiDAR motion considering the dynamic characteristics of LiDAR point clouds. Finally, the proposed method results in a global map with static point clouds and accurate LiDAR motion with global pose optimization. We tested the proposed method on the well-known public dataset (KITTI) and the custom dataset with complex environments, including various moving objects. Comparisons with state-of-the-art (SOTA) methods demonstrate that the our approach is more robust and efficient. For example, the proposed method performed an average 0.63% and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.18^{\circ }/100\;m$ </tex-math></inline-formula> errors on the KITTI dataset with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.96ms$ </tex-math></inline-formula> processing time that convinces real-time processing.

Relative Pose Estimation of Non-Cooperative Space Targets Using a TOF Camera

It is difficult to determine the accurate pose of non-cooperative space targets in on-orbit servicing (OOS). The visual camera is easily affected by the extreme light environment in space, and the scanning lidar will have motion distortion when the target moves at high speed. Therefore, we proposed a non-cooperative target pose-estimation system combining a registration and a mapping algorithm using a TOF camera. We first introduce the projection model of the TOF camera and proposed a new calibration method. Then, we introduce the three modules of the proposed method: the TOF data preprocessing module, the registration module and the model mapping module. We assembled the experimental platform to conduct semi-physical experiments; the results showed that the proposed method has the smallest translation error 8 mm and Euler angle error 1° compared with other classical methods. The total time consumption is about 100 ms, and the pose tracking frequency can reach 10 Hz. We can conclude that the proposed pose-estimation scheme can achieve the high-precision pose estimation of non-cooperative targets and meet the requirements necessary for aerospace applications.

DASFNet: Dense-Attention–Similarity-Fusion Network for scene classification of dual-modal remote-sensing images

• The cosine similarity and a gating mechanism are used by the SFM to fuse IRRG and DSM data and extract corresponding complementary features. • The MCFAM is used to extract multiscale context cues, solving the problem of object-scale variations in HRRSIs. • The DAMM forms a dense-attention branch that can adaptively capture both low- and high-level attentions. Although significant progress has been made in scene classification of high-resolution remote-sensing images (HRRSIs), dual-modal HRRSI scene classification is still an active and challenging issue. In this study, we introduce an end-to-end dense-attention–similarity-fusion network (DASFNet) for dual-modal HRRSIs. Specifically, we propose a dense-attention map module based on graph convolution, which adaptively captures long-range semantic cues and further directs shallow-attention cues to the deep level to guide the generation of high-level feature attention cues. At the encoder stage, DASFNet uses feature similarity to explore the correlation between dual-modal features; a similarity-fusion module extracts complementary information by fusing features from different modalities. A multiscale context-feature-aggregation module is used to strengthen the feature embedding of any two spatial scales; this solves the of scale change problem. A large number of experiments on two HRRSI benchmark datasets for scene classification indicate that the proposed DASFNet outperforms the outstanding scene classification approaches.

Classification of hyperchaotic, chaotic, and regular signals using single nonlinear node delay-based reservoir computers.

The Lyapunov exponent method is generally used for classifying hyperchaotic, chaotic, and regular dynamics based on the equations modeling the system. However, several systems do not benefit from appropriate modeling underlying their dynamic behaviors. Therefore, having methods for classifying hyperchaotic, chaotic, and regular dynamics using only the observational data generated either by the theoretical or the experimental systems is crucial. In this paper, we use single nonlinear node delay-based reservoir computers to separate hyperchaotic, chaotic, and regular dynamics. We show that their classification capabilities are robust with an accuracy of up to 99.61% and 99.03% using the Mackey-Glass and the optoelectronic oscillator delay-based reservoir computers, respectively. Moreover, we demonstrate that the reservoir computers trained with the two-dimensional Hénon-logistic map can classify the dynamical state of another system (for instance, the two-dimensional sine-logistic modulation map). Our solution extends the state-of-the-art machine learning and deep learning approaches for chaos detection by introducing the detection of hyperchaotic signals.

ПРОСТРАНСТВЕННО-ВРЕМЕННАЯ ИЗМЕНЧИВОСТЬ ПОВЕРХНОСТНОГО СТОКА РЕК БЕЛАРУСИ

An assessment of changes in the surface river runoff of Belarus in modern conditions is given. Significant negative trends were found for the series of the percentage of surface runoff in the total volume of annual runoff for the period 1948-2017. The greatest decrease in surface runoff is observed in the south of Belarus. The degree of synchronicity of fluctuations of the long-term surface runoff of large rivers of Belarus is estimated. A modern map of the average annual modules of the surface river runoff distribution in Belarus has been obtained.

AMFFNet: attention-guided multi-level feature fusion network for land cover classification of remote sensing images

In the field of remote sensing, the classification of land cover is a pivotal and challenging issue. Standard models fail to capture global and semantic information in remote sensing images despite the fact that a convolutional neural network provides robust support for semantic segmentation. In addition, owing to disparities in semantic levels and spatial resolution, the simple fusion of low-level and high-level features may diminish the efficiency. To address these deficiencies, an attention-guided multi-level feature fusion network (AMFFNet) is proposed in this study. The proposed AMFFNet approach is designed as an encoder–decoder network with the inclusion of a multi-level feature fusion module (MFF) and a dual attention map module (DAM). A DAM models the semantic association of features from a spatial and channel perspective, and an MFF bridges the semantic and resolution gaps between high-level and low-level features. Furthermore, we propose a residual-based boundary refinement upsample module to further optimize the object boundaries. The experimental results indicate that the proposed strategy can considerably enhance the accuracy of land cover classification, achieving a mean intersection over union of 90.39% on the LandCover.ai dataset and 63.14% on the Gaofen Image Dataset with 15 categories (GID-15).

Research on the autonomous system of the quadruped robot with a manipulator to realize leader‐following, object recognition, navigation and operation

AbstractA systematic solution is developed to improve the autonomous capability of the quadruped robot with a manipulator, such as navigation, recognition and operation. The developed system adopts novel software, hardware system and system architecture, including a specially designed environment awareness system (EAS). Based on the camera and LiDAR on the EAS, the recognition of multiple common targets, such as the leader, door, window and bag, is achieved. In terms of navigation, a location method is built, that combines the laser odometer and global positioning system. A mapping and path planning module is designed by the Robot‐centric Elevation Mapping algorithm and the rapidly exploring rand tree algorithm. For operation, a real‐time target grasp detection system is proposed based on the You Only Look Once v5 algorithm to improve the success rate of tasks. The whole system is integrated based on the task relevance scheduling strategy to reduce the computational complexity. The tightly integrated system and the subsystems are evaluated by conducting simulations and physical experiments in robot recognition, navigation and operation. Extensive experiments show that the proposed framework can better achieve the autonomous navigation and operation of the quadruped robot with a manipulator. Notably, the proposed framework is still effective when facing dynamic objects. In addition, the system can be easily extended to other forms of mobile robot.

FlowTracker: Improved flow correlation attacks with denoising and contrastive learning

Tor is the most widely used anonymous network which provides anonymity by using proxy nodes to route data. However, many researches have proved that flow correlation attacks can break the anonymity of users’ communication relationships: once attackers have access to the ingress and egress traffic flows at both ends of the network, they will find the specific user who is visiting a monitored Internet service by observing the traffic shape. Existing flow correlation attacks have limited effects because of insufficient traffic shape features, network noise influence and lack of differentiated learning on unrelated flows. In this paper, we propose an improved flow correlation attacking model named FlowTracker which contains four modules: In flow cumulative representation module, raw and normalized cumulative sequences are combined to construct robust traffic shape features. In noise resistant mapping module, stacked auto-encoders are adopted to learn the mapping relationship between related ingress and egress cumulative representations and filter out the network noise. Moreover, we leverage contrast learning in distance optimization module to generate the optimized representation in which difference between unrelated representations is amplified, further reducing the false correlation rate with similar but unrelated flows. Finally, the target flow is reported through a multi-layer decision process in the hierarchical judgment module. We evaluate FlowTracker and other advanced comparison methods on the public and self-built datasets with the more realistic unidirectional setting, experiment results demonstrate that FlowTracker has the highest attacking success rate under general and targeted scenarios, especially when the traffic shape is seriously destroyed by traffic obfuscation.

Unpaired referring expression grounding via bidirectional cross-modal matching

Referring expression grounding is an important and challenging task in computer vision. To avoid the laborious annotation in conventional referring grounding, unpaired referring grounding is introduced, where the training data only contains a number of images and queries without correspondences. The few existing solutions to unpaired referring grounding are still preliminary, due to the challenges of learning vision-language correlation and lack of the top-down guidance with unpaired data. Existing works are only able to learn vision-language correlation by modality conversion, where critical information are lost. They also heavily rely on pre-extracted object proposals and thus cannot generate correct predictions with defective proposals.In this paper, we propose a novel bidirectional cross-modal matching (BiCM) framework to address these challenges. Particularly, we design a query-aware attention map (QAM) module that introduces top-down perspective via generating query-specific visual attention maps to avoid the over-reliance on pre-extracted object proposals. A cross-modal object matching (COM) module is further introduced to predict the target objects from a bottom-up perspective. This module exploits the recently emerged image-text matching pretrained model, CLIP, to learn cross-modal correlation without modality conversion. The top-down and bottom-up predictions are then integrated via a similarity fusion (SF) module. We also propose a knowledge adaptation matching (KAM) module that leverages unpaired training data to adapt pretrained knowledge to the target dataset and task. Experiments show that our framework significantly outperforms previous works on five grounding datasets.

Modulation and Direct Mapping of the Interfacial Band Alignment of an Eco-Friendly Zinc-Tin-Oxide Buffer Layer in SnS Solar Cells

Modulation and Direct Mapping of the Interfacial Band Alignment of an Eco-Friendly Zinc-Tin-Oxide Buffer Layer in SnS Solar Cells

Clinical application of digitalization of occlusal contacts with dental scanner

This study aimed to evaluate the number, intensity and position of occlusal contact points in a case of mesiocclusion, hyperdivergence, with open bite. An occlusal record was taken from a patient with anterior and lateral open bite mesiocclusion, using the Planmeca Esmerald S intraoral scanner in maximum intercuspation. The intensity of the occlusal contact was analyzed with the software 3shape Ortho Analyzer Orthodontics, using the Occlusion Map module, through the 3D Color Map tool, with a 0.5 mm virtual articular paper. These results were compared to the occlusal support points defined by Planas [10]. The interpretation of the data obtained was made by assessing the interocclusal intensity of the contact points, number of contacts and position during three different moments (1S, 2S, 3S) in the record taking process. The chromatic scale of the Color Map is: red, orange, yellow, green and blue. To identify the occlusal contact points in digital, they are shown in red points when full contact occurs, while minimum contact is shown in blue. We evaluated the number of teeth with interocclusal contacts. It was determined that having the appropriate number of contacts does not imply that they are in the correct position. In addition, the method suggests reliability in the filing and record keeping of occlusal contacts. By identifying intensity, number and position of the occlusal support point we can objectively record interocclusal alterations.

An Efficient Chaos-Based Image Encryption Technique Using Bitplane Decay and Genetic Operators.

Social networks have greatly expanded in the last ten years the need for sharing multimedia data. However, on open networks such as the Internet, where security is frequently compromised, it is simple for eavesdroppers to approach the actual contents without much difficulty. Researchers have created a variety of encryption methods to strengthen the security of this transmission and make it difficult for eavesdroppers to get genuine data. However, these conventional approaches increase computing costs and communication overhead and do not offer protection against fresh threats. The problems with current algorithms encourage academics to further investigate the subject and suggest new algorithms that are more effective than current methods, that reduce overhead, and which are equipped with features needed by next-generation multimedia networks. In this paper, a genetic operator-based encryption method for multimedia security is proposed. It has been noted that the proposed algorithm produces improved key strength results. The investigations using attacks on data loss, differential assaults, statistical attacks, and brute force attacks show that the encryption technique suggested has improved security performance. It focuses on two techniques, bitplane slicing and followed by block segmentation and scrambling. The suggested method first divides the plaintext picture into several blocks, which is then followed by block swapping done by the genetic operator used to combine the genetic information of two different images to generate new offspring. The key stream is produced from an iterative chaotic map with infinite collapse (ICMIC). Based on a close-loop modulation coupling (CMC) approach, a three-dimensional hyperchaotic ICMIC modulation map is proposed. By using a hybrid model of multidirectional circular permutation with this map, a brand-new colour image encryption algorithm is created. In this approach, a multidirectional circular permutation is used to disrupt the image’s pixel placements, and genetic operations are used to replace the pixel values. According to simulation findings and security research, the technique can fend off brute-force, statistical, differential, known-plaintext, and chosen-plaintext assaults, and has a strong key sensitivity.

A map of cis-regulatory modules and constituent transcription factor binding sites in 80% of the mouse genome

BackgroundMouse is probably the most important model organism to study mammal biology and human diseases. A better understanding of the mouse genome will help understand the human genome, biology and diseases. However, despite the recent progress, the characterization of the regulatory sequences in the mouse genome is still far from complete, limiting its use to understand the regulatory sequences in the human genome.ResultsHere, by integrating binding peaks in ~ 9,000 transcription factor (TF) ChIP-seq datasets that cover 79.9% of the mouse mappable genome using an efficient pipeline, we were able to partition these binding peak-covered genome regions into a cis-regulatory module (CRM) candidate (CRMC) set and a non-CRMC set. The CRMCs contain 912,197 putative CRMs and 38,554,729 TF binding sites (TFBSs) islands, covering 55.5% and 24.4% of the mappable genome, respectively. The CRMCs tend to be under strong evolutionary constraints, indicating that they are likely cis-regulatory; while the non-CRMCs are largely selectively neutral, indicating that they are unlikely cis-regulatory. Based on evolutionary profiles of the genome positions, we further estimated that 63.8% and 27.4% of the mouse genome might code for CRMs and TFBSs, respectively.ConclusionsValidation using experimental data suggests that at least most of the CRMCs are authentic. Thus, this unprecedentedly comprehensive map of CRMs and TFBSs can be a good resource to guide experimental studies of regulatory genomes in mice and humans.