Rectification Method Research Articles

A Two-Level Rectification Attention Network for Scene Text Recognition

Scene text recognition is a challenging task in the computer vision field due to the diversity of text styles and the complexity of the image backgrounds. In recent decades, numerous text rectification and recognition methods have been proposed to solve these problems. However, most of these methods rectify texts at the geometry level or pixel level. The former is limited by geometric constraints, and the latter is prone to blurring the text. In this paper, we propose a two-level rectification attention network (TRAN) to rectify and recognize texts. This network consists of two parts: a two-level rectification network (TORN) and an attention-based recognition network (ABRN). Specifically, the TORN first rectifies texts at the geometry level and then performs a pixel-level adjustment, which not only eliminates the geometric constraints but also renders clear texts. The ABRN's role is to recognize text in the rectified images. To improve the feature extraction ability of our model, we design a new channel-wise and kernel-wise attention unit, which enables the network to handle significant variations of character size and channel interdependencies. Furthermore, we propose a skip training strategy to make our model converge smoothly. We conduct experiments on various benchmarks, including regular and irregular datasets. The experimental results show that our method achieves a state-of-the-art performance.

SIR: Self-supervised Image Rectification via Seeing the Same Scene from Multiple Different Lenses.

Deep learning has demonstrated its power in image rectification by leveraging the representation capacity of deep neural networks via supervised training based on a large-scale synthetic dataset. However, the model may overfit the synthetic images and generalize not well on real-world fisheye images due to the limited universality of a specific distortion model and the lack of explicitly modeling the distortion and rectification process. In this paper, we propose a novel self-supervised image rectification (SIR) method based on an important insight that the rectified results of distorted images of a same scene from different lenses should be the same. Specifically, we devise a new network architecture with a shared encoder and several prediction heads, each of which predicts the distortion parameter of a specific distortion model. We further leverage a differentiable warping module to generate the rectified images and re-distorted images from the distortion parameters and exploit the intra- and inter-model consistency between them during training, thereby leading to a self-supervised learning scheme without the need for ground-truth distortion parameters or normal images. Experiments on synthetic dataset and real-world fisheye images demonstrate that our method achieves comparable or even better performance than the supervised baseline method and representative state-of-the-art (SOTA) methods. The proposed self-supervised method also provides a possible way to improve the universality of distortion models while keeping their self-consistency. Code and datasets will be available at https://github.com/loong8888/SIR.

Accurate Head Pose Estimation Using Image Rectification and a Lightweight Convolutional Neural Network

Head pose estimation is an important step for many human-computer interaction applications such as face detection, facial recognition, and facial expression classification. Accurate head pose estimation benefits these applications that require face images as the input. Most head pose estimation methods suffer from perspective distortion because the users do not always align their face perfectly with the camera. This paper presents a new approach that uses image rectification to reduce the negative effect of perspective distortion and a lightweight convolutional neural network to obtain highly accurate head pose estimations. The proposed method calculates the angle between the optical axis of the camera and the projection vector of the center of the face. The face image is rectified using this estimated angle through perspective transformation. A lightweight network that is only 0.88 MB in size is designed to take the rectified face image as the input to perform head pose estimation. The output of the network, the head pose estimation of the rectified face image, is transformed back to the camera coordinate system as the final head pose estimation. Experiments on public benchmark datasets show that the proposed image rectification method and the newly designed lightweight network improve the accuracy of head pose estimation remarkably. Compared with state-of-the-art methods, our approach achieves both higher accuracy and faster processing speed.

Optimized rectification of fiber orientation density function with background threshold

PurposeTo describe an optimized fiber orientation density function (fODF) rectification procedure that removes negative values and absorbs all features below a specified threshold into a constant background. Theory and methodsThe fODF for a white matter imaging voxel describes the angular density of axons. Because of signal noise and Gibbs ringing, fODFs estimated with diffusion MRI may take on unphysical negative values in some directions and contain spurious peaks. In order to suppress such artifacts, an fODF rectification procedure is proposed that both eliminates all negative values and incorporates all features below a specified threshold, η, into a constant background while at the same time minimizing the mean square deviation from the original, unrectified fODF. Calculating this fODF is straightforward, and the directions and shapes of peaks not absorbed into the background are preserved. The rectification method is illustrated for an analytic fODF model and for experimental diffusion MRI data obtained in healthy human brain, with the original fODFs being obtained from fiber ball imaging. ResultsExamples of optimal rectified fODFs are given for three choices of the background threshold referred to as minimal rectification (η = 0), average-level rectification (η ≈ 0.08), and fractional-anisotropy-axonal-based rectification (η ≈ 0.1). As η is increased, artifacts and other small features are more strongly suppressed, but the major fODF peaks are largely unaffected for the range of η values illustrated by these three alternatives. ConclusionArtifactual features of fODFs estimated with diffusion MRI can be effectively suppressed by applying the proposed optimized rectification procedure. Since it minimizes fODF distortion in the mean square sense, it may be useful in the study of how fODF fine structure is affected by aging and disease.

Control of Free Piston Stirling Linear Generator system connected with dc/dc converter for energy storage applications based on SVPWM Rectification Method

Recently, the world has been facing several challenges in terms of environmental change, which is expected to increase in the coming periods. Further challenging is how to reduce the emissions of greenhouse gasses into the atmosphere. Such emissions pollute our atmosphere, resulting in global warming. However, the wide use of renewable energies can both help to fight against this phenomenon and reach long-term goals. Considering the growing global concern over the use of fossil fuels and environmental pollution. In this case, a Stirling engine with diversified heat sources and low pollution was developed, and the Free Piston Stirling Linear Generator (FPSLG), which can be directly connected to the Stirling engine and thus eliminate the need for an intermediate mechanical transmission mechanism, was used to create a power generation system with the Stirling engine and has been employed in a range of applications, including space power supply, solar energy generation, and others. This article models two sets of power converters, one of which is connected to the FPSLG and the other to the storage battery system. The Space Vector Pulse Width Modulation (SVPWM) control methods adaption to the AC/DC rectifier and the two-loop Proportional Integral (PI) control method adaption to the DC/DC converter used for renewable energy conversion is proposed. Furthermore, the traditional PI in the position loop is replaced by a Proportional Resonance (PR) controller to increase the frequency and displacement tracking performance. The linear generator’s control system is simulated, and the efficiency of the proposed control strategy is proven. The system was initially simulated in MATLAB/Simulink using the dq-PI current controller to produce the logic pattern for the Voltage Source Rectifier (VSR), and the resulting logic pattern was examined.

Terahertz generation optimization in an OH1 nonlinear organic crystal pumped by a Cr:forsterite laser.

We present the results of experimental investigations of terahertz radiation generation conversion efficiency in an OH1 nonlinear organic crystal pumped by femtosecond laser pulses at 1240 nm wavelength. An influence of OH1 crystal thickness on the terahertz generation by optical rectification method was studied. It is shown that the optimal crystal thickness for the maximum conversion efficiency is 1 mm, which agrees with the previously made theoretical estimates.

Exploring density rectification and domain adaption method for crowd counting.

Crowd counting has received increasing attention due to its important roles in multiple fields, such as social security, commercial applications, epidemic prevention and control. To this end, we explore two critical issues that seriously affect the performance of crowd counting including nonuniform crowd density distribution and cross-domain problems. Aiming at the nonuniform crowd density distribution issue, we propose a density rectifying network (DRNet) that consists of several dual-layer pyramid fusion modules (DPFM) and a density rectification map (DRmap) auxiliary learning module. The proposed DPFM is embedded into DRNet to integrate multi-scale crowd density features through dual-layer pyramid fusion. The devised DRmap auxiliary learning module further rectifies the incorrect crowd density estimation by adaptively weighting the initial crowd density maps. With respect to the cross-domain issue, we develop a domain adaptation method of randomly cutting mixed dual-domain images, which learns domain-invariance features and decreases the domain gap between the source domain and the target domain from global and local perspectives. Experimental results indicate that the devised DRNet achieves the best mean absolute error (MAE) and competitive mean squared error (MSE) compared with other excellent methods on four benchmark datasets. Additionally, a series of cross-domain experiments are conducted to demonstrate the effectiveness of the proposed domain adaption method. Significantly, when the A and B parts of the Shanghaitech dataset are the source domain and target domain respectively, the proposed domain adaption method decreases the MAE of DRNet by 47.6%.

Comparative Analysis of Riser Base and Flowline Gas Injection on Vertical Gas-Liquid Two-Phase Flow

Gas injection is a frequently used method for artificial lift and flow regime rectification in offshore production and transportation flowlines. The flow behaviour in such flowlines is complex and a better understanding of flow characteristics, such as flow patterns, void fraction/hold up distributions and pressure gradient is always required for efficient and optimal design of downstream handling facilities. Injection method and location have been shown to strongly affect downstream fluid behaviour that can have important implications for pumping and downstream facility design, especially if the development length between pipeline and downstream facility is less than L/D = 50 as reported by many investigators. In this article, we provide the results of an experimental investigation into the effects of the gas injection position on the characteristics of the downstream upwards vertical gas flow using a vertical riser with an internal diameter of 52 mm and a length of 10.5 m. A horizontal 40-m-long section connected at the bottom provides options for riser base or horizontal flow line injection of gas. The flowline gas injection is performed 40 m upstream of the riser base. A 16 by 16 capacitance wire mesh sensor and a gamma densitometer were used to measure the gas-liquid phase cross-sectional distribution at the riser top. A detailed analysis of the flow characteristics is carried out based on the measurements. These include flow regimes, cross-sectional liquid holdup distributions and peaking patterns as well as analysis of the time series data. Our findings show that flow behaviours differences due to different gas injection locations were persisting after a development length of 180D in the riser. More specifically, core-peaking liquid holdup occurred at the lower gas injection rates through the flowline, while wall-peaking holdup profiles were established at the same flow conditions for riser base injection. Wall peaking was associated with dispersed bubbly flows and hence non-pulsating as against core-peaking was associated with Taylor bubbles and slug flows. Furthermore, it was found that the riser base injection generally produced lower holdups. It was noted that the circumferential injector used at the riser base promoted high void fraction and hence low liquid holdups. Due to the bubbly flow structure, the slip velocity is smaller than for larger cap bubbles and hence the void fraction is higher. The measurements and observations presented in the paper provides valuable knowledge on riser base/flowline gas introduction that can directly feed into the design of downstream facilities such as storage tanks, slug catchers and separators.

Novel Synchronous Rectification and Receiving Power Compensation Method for WPT Only by DC Current Sensor

Dynamic wireless power transfer can solve the shorter cruising rangeproblem of electric vehicles. The synchronous rectification can improve the efficiency of wireless power transfer, but it costs more because it requires an expensive ac current sensor. A novel synchronous rectification method is proposed which only uses an inexpensive dc current sensor and does not measure ac current or voltage. This method continuously searches the synchronous state by the gradient descent method and the feedforward phase jump control. This method also compensates for the decreased received power caused by the deviation of the parameters. In the first part, it is explained by the equivalent circuit model that the proposed method can achieve synchronous rectification by maximizing the dc received current. Then, the effectiveness of the proposed method in terms of the efficiency and the received power is verified in the experiment by comparing it with the conventional diode rectification. The proposed method reduced the loss by 31.77% and achieved synchronous rectification four times faster.

Shell DFT-1/2 method towards engineering accuracy for semiconductors: GGA versus LDA

The Kohn-Sham gaps of density functional theory (DFT) obtained in terms of local density approximation (LDA) or generalized gradient approximation (GGA) cannot be directly linked to the fundamental gaps of semiconductors, but in engineering there is a strong demand to match them through certain rectification methods. Shell DFT-1/2 (shDFT-1/2), as a variant of DFT-1/2, is a potential candidate to yield much improved band gaps for covalent semiconductors, but its accuracy depends on the LDA/GGA ground state, including optimized lattice parameters, basic Kohn-Sham gap before self-energy correction and the amount of self-energy correction that is specific to the exchange-correlation (XC) functional. In this work, the LDA/GGA as well as shDFT-1/2 results of six technically important covalent semiconductors Si, Ge, GaN, GaP, GaAs and GaSb were tested, with an additional ionic insulator LiF for comparison. The impact of XC flavor (LDA, PBEsol, PBE and RPBE), either directly on the gap value, or indirectly through the optimized lattice constant, is examined comprehensively. Moreover, the impact of XC flavor on LDA/GGA and shDFT-1/2 gaps was investigated under the condition of fixed experimental lattice constants. In-depth analysis reveals the rule of reaching the best accuracy in calculating the electronic band structures of typical covalent semiconductors. Relevant parameters like lattice constant, self-consistency in shDFT-1/2 runs, as well as the exchange enhancement factor of GGA, are discussed in details.

Horizontal Accuracy Assessment of Rectified Map, Case Study: Map of Boundary Areas of Konak Forest at Rejang Lebong Regency, Bengkulu Province

Spatial archives in the form of analog maps still encountered in several work units, however, as times progressed, the quality of these maps decreased in terms of the materials used which could affect their accuracy. There are several techniques used in converting from analog to digital maps, including scanning technique, and a unification coordinate system with the rectification method. This study aimed to find out the horizontal accuracy and the accuracy class of rectified maps. The data used was an analog map of a map of the boundary demarcation of the Konak Forest area, Rejang Lebong Regency scale 1:1000. The analog map was produced more than 30 years ago. The steps carried out in this study were preparing data from scanning analog maps, setting coordinate systems, input GCP, control rectification, calculating RMSE and CE values, analyzing, digitizing, layout. The result of the rectification map shows the use of 4GCP produces the best value with RMSE 0.14m and CE90 0,20 and the horizontal accuracy is class 1 on a scale of 1:1000. 6GCP produces RMSE 0.5m and CE90 0.75 and the horizontal accuracy is class 3 on a scale of 1:2500. 10GCP produces RMSE 0.6m and CE90 0.95 and the horizontal accuracy is class 3 on a scale of 1:2500.

Stereo-phase rectification for metric profilometry with two calibrated cameras and one uncalibrated projector.

Fringe projection profilometry requires calibrating both cameras and projectors for metric measurements. Cameras are relatively simple to calibrate, but projectors require more sophisticated procedures. In this paper, a fringe projection profilometer with two calibrated cameras and one uncalibrated projector is developed for metric measurements. A phase rectification method, which is crucial for stereo matching, is designed by minimizing the perspective distortion. Also, a simple method for point matching using stereo rectified phase maps is proposed. The principles of metric profilometry using the proposed rectification method are introduced. The developed system is evaluated experimentally by the metric measurement of three-dimensional objects. The obtained results confirm a high accuracy of metric measurement and versatility in the design of fringe projection profilometers with uncalibrated projectors.

Discriminative error prediction network for semi-supervised colon gland segmentation.

Pixel-wise error correction of initial segmentation results provides an effective way for quality improvement. The additional error segmentation network learns to identify correct predictions and incorrect ones. The performance on error segmentation directly affects the accuracy on the test set and the subsequent self-training with the error-corrected pseudo labels. In this paper, we propose a novel label rectification method based on error correction, namely ECLR, which can be directly added after the fully-supervised segmentation framework. Moreover, it can be used to guide the semi-supervised learning (SSL) process, constituting an error correction guided SSL framework, called ECGSSL. Specifically, we analyze the types and causes of segmentation error, and divide it into intra-class error and inter-class error caused by intra-class inconsistency and inter-class similarity problems in segmentation, respectively. Further, we propose a collaborative multi-task discriminative error prediction network (DEP-Net) to highlight two error types. For better training of DEP-Net, we propose specific mask degradation methods representing typical segmentation errors. Under the fully-supervised regime, the pre-trained DEP-Net is used to directly rectify the initial segmentation results of the test set. While, under the semi-supervised regime, a dual error correction method is proposed for unlabeled data to obtain more reliable network re-training. Our method is easy to apply to different segmentation models. Extensive experiments on gland segmentation verify that ECLR yields substantial improvements based on initial segmentation predictions. ECGSSL shows consistent improvements over a supervised baseline learned only from labeled data and achieves competitive performance compared with other popular semi-supervised methods.

Near-automated orthorectification of archived aerial photographs

The proposed objective was to develop a near-automated workflow for orthorectification of archived aerial photographs that overcomes the limited resources problem. Automation was implemented using the oriented fast and rotated brief (ORB) and random sample consensus (RANSAC) for conjugate points detection. Because it is difficult to determine the RANSAC threshold, the RANSAC threshold search for two step triangulation method was developed to find the threshold that resulted in the minimum three-dimensional misclosures at check points from the threshold candidates. For triangulation, the bundle block adjustment (BBA) with a minimal constraint and the absolute orientation were implemented. Although the absolute orientation still requires manual measurement of control points, the required number is usually less than that of traditional BBA, especially for longer strips. The control points were measured in relatively recent high-accuracy orthophotos and the digital elevation model. In the experiments, the archived aerial strip photographs were orthorectified using a differential rectification method with ∼1 m resolution. The first and second orthorectification experiments showed 1.31 m (X), 1.14 m (Y), and 1.88 m (X), 1.51 m (Y) of RMSE. The proposed workflow suggested is significant because it can save cost, time, and labor required for the orthorectification of archived aerial photographs while achieving reasonable accuracy.

Keyframe-Editable Real-Time Motion Synthesis

Since existing motion synthesis methods often lack precise controls to the synthesis process, we propose a keyframe-editable motion synthesis framework which allows users to edit the keyframes of an expected motion sequence and use the edited keyframes to control and drive the synthesis process. Specifically, a motion segment can be represented as a start-end frame pair and a group of motion patterns which record the joint angle changing processes between the start and end frames. With the pre-trained paired dictionaries relating the start-end frame space and motion pattern space, we can use the edited keyframe pair to generate its corresponding motion pattern, so the validity of keyframes is very important to the quality of synthesized motions. Thus, we use the probability to measure the motion naturalness and propose a naturalness rectification method to guarantee the validity of the edited keyframes. We also provide a joint-move interface and propose a position refinement method for the detailed adjustment of the joint positions of keyframes. Besides, we add extra naturalness constraint to the motion synthesis process to further improve the naturalness of the generated in- between frames. Extensive rectification experiments in different situations verify the effect of the proposed naturalness rectification model. The comparisons of the synthesis results with other state-of-the-art synthesis methods demonstrate the advantage of our motion synthesis model. The high efficiencies of all the algorithms make the proposed framework competent to the real-time motion synthesis.

Backward Step-Up Control Strategy for Bidirectional LLC Resonant Converter

An LLC resonant converter has the advantages of simple structure and soft switching. It can enable bidirectional power transmission, but it is difficult to realize a normalized gain greater than one under backward mode (backward step-up mode). Cascaded dc/dc converters or topological changes can solve this problem, but additional switches and components are required and losses are added. Without changing the LLC resonant converter’s basic topology, this paper proposes a variable duty-cycle control strategy of primary side switches for backward step-up mode. Using variable duty-cycle control, the LC resonant tank can be charged, and then the backward step-up mode can be realized. Soft switching characteristics of some primary side switches and all secondary side switches are guaranteed. In this study, the working principle of an LLC resonant converter under bidirectional control strategy was analyzed, and the backward step-up control was analyzed in detail. The voltage gain and the boundary of continuous conduction mode (CCM) and discontinuous conduction mode (DCM) were derived. A synchronous rectification method related to the backward step-up control is proposed. The control strategy was verified by experiments.

Binocular Images Dense Matching considering Image Adaptive Color Weights and Feature Points

When the matching cost function in Semiglobal Matching is unstable, the inaccurate matching cost values will be propagated in the cost aggregation process. It will lead to a serious mismatching phenomenon. To address the problem, a binocular images dense matching method considering image adaptive color weights and feature points was proposed. Firstly, The Color Birchfield Tomasi (CBT) matching cost calculation method was proposed to obtain a stable initial cost volume, which combined image adaptive color weights and gradient information. Secondly, the Scale-invariant Feature Transform matching algorithm was used to extract the a priori feature points from binocular images. Then, the feature points were filtrated. The cost volume was optimized by using their coordinate information and disparity information. Finally, an aggregation path segmentation rectification method was adopted to optimize the SGM aggregation paths and reduce the propagation of incorrect paths. Experimental results demonstrate that the proposed method can effectively improve the stability and accuracy of dense matching, reduce the mismatching phenomenon, and finally produce high-quality disparity maps.

An H5-Bridge-Based Laddered $CLLC$ DCX With Variable DC Link for PEV Charging Applications

Plug-in electric vehicles’ charger is preferred to cover an ultrawide battery voltage range with the vehicle-to-grid capability. Conventional bidirectional resonant dc–dc converters suffer from the contradiction among wide voltage gain range, squeezed dc-link voltage span, and narrow switching frequency band. To solve the issue, this article proposes a novel H5-bridge-based bidirectional <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$CLLC$</tex-math></inline-formula> converter. By configuring the switch pattern, the H5-bridge can form the modes of single half-bridge, dual half-bridge, half full-bridge, and dual full-bridge, respectively. Correspondingly, six gain curves can be derived. Combined with the variable dc-link framework, the converter constrains the switching frequency in the vicinity of the resonant frequency with optimal efficiency. The converter achieves an ultrawide battery voltage range with a squeezed dc-link span. A bidirectionally synchronous rectification method is proposed to improve the efficiency further. To verify the proposed concept, a 1-kW rated prototype with a 320–420 V dc link is built and tested. It validates the battery voltage 55–420 V for charging and 230–420 V for discharging. Zero-voltage turn- <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">on</small> and zero-current turn- <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">off</small> are achieved in the rectifying <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">mosfet</small> s. The prototype exhibits 98.04% peak efficiency and good overall efficiency performance.

A Current Self-Injection Based Position Sensorless Control for DSEG With Controlled Rectification

The controlled rectification method with positon sensor can effectively improve the output power of the doubly salient electromagnetic generator (DSEG), but weakens the advantage of no position sensor for the traditional DSEG with uncontrolled rectification. To solve this problem, a current self-injection based position sensorless control method is proposed for DSEG with H-bridge semi-controlled rectifier (HBSCR) in this article. Based on the operation principle of DSEG and HBSCR, the position estimation method is developed from the positive zero-crossing point of inductive electromotive force (EMF), which is indirectly observed through the test current self-injected by the inductive EMF. According to the relationship between generation current and test current, a test current self-injection method without presetting injection width is put forward, and the corresponding sensorless control strategy is established with the combination of angular position semi-control. Then, the compensation strategy is further presented to improve the accuracy of the proposed position estimation method. Finally, the simulations and experiments are implemented on a three-phase DSEG to validate the feasibility and correctness of the proposed sensorless control strategy.

Improved variable scale-convex-peak method for weak signal detection

Based on the rectification method of detectable domain and elimination method of blind domain, the variable scale-convex-peak method for the identification frequency of weak signal is improved. For the initial phase of weak signal to be detected, the analytic expression of chaotic threshold is obtained by using the stochastic Melnikov method. The influence of phase on chaotic threshold is clarified, which leads to that the frequency of weak signal can be identified. Specifically, within a period, the initial phase can be divided into the detectable domain and the blind domain. In the detectable domain, when the frequency identification deviates, the rectification method of detectable domain is introduced. Specifically, the Newton interpolation method is used to give the calculation formula to rectify the deviation, which improves the accuracy of frequency identification. In the blind domain, the elimination method of blind domain is given by constructing the detection equations. The feasibility of the improved variable scale-convex-peak method is verified by numerical simulation and circuit experiment. Furthermore, by comparing the identified frequency value with the theoretical fault frequency value, the fault location of the wheelset bearing of the high-speed train is determined. The improved variable scale-convex-peak method is applied when the initial phase is in the frequency detectable domain or the blind domain, respectively.