Centerline Extraction Algorithm Research Articles

An enhanced centerline extraction algorithm for complex stripes in linear laser scanning measurement

Centerline extraction is a basic and critical step for linear laser scanning measurement. However, in practical measurement, the inconsistent reflectivity of the measured surface and the external noise interference can result in complex variation of the stripe pattern, which will influence the centerline extraction accuracy of the stripe pattern and hence the measurement accuracy of the laser scanner. To solve this problem, an enhanced centerline extraction algorithm for complex laser stripes is proposed. In the preprocessing of the stripe pattern, a region separation strategy is employed to mitigate the effects of complex variations, and multi-region self-adaptive convolution is adopted to enhance the stripe pattern quality. For high-precision extraction of the stripe centerline, different convolution kernels are applied to compute the Hessian matrix for the stripe patterns in different regions to determine the normal direction of the laser line. The second order Taylor expansion is performed along this direction to work with the denoising algorithm to determine the subpixel positions of the center points on the line, and then the whole centerline of the stripe pattern is obtained by piecewise cubic Hermite interpolation methods. Experimental results show that the effectiveness of the proposed algorithm in addressing centerline extraction for complex laser stripe patterns due to stray light interference, reflective stripes, and laser stripe linewidth changes in laser scanning measurement.

Rapid measurement method for cable tension of cable‐stayed bridges using terrestrial laser scanning

AbstractThis study proposes a new method for the rapid and non‐contact measurement of cable forces in cable‐stayed bridges, including a cable force calculation method based on measured cable shapes and a batch acquisition method for the true shape of cables. First, a nonlinear regression model for estimating cable forces based on measured cable shapes is established, and a Gauss–Newton‐based cable force solving method is proposed. Furthermore, terrestrial laser scanning technology is used to collect geometric data of the cables. Meanwhile, automatic segmentation, noise reduction, and centerline extraction algorithms for the cable point cloud are proposed to accurately and efficiently obtain the cable shape. The correctness of the proposed cable force calculation method is verified in a well‐designed experiment, with the measurement error of cable forces for 15 test samples being less than 1%. Finally, the proposed point cloud automation processing algorithm and cable force measurement method are fully tested on a cable‐stayed bridge with a span of 460 m. The measurement accuracy of the proposed method for actual bridge cable tension is comparable to that of the frequency method, but the detection efficiency on site is nine times higher than that of the traditional frequency method. Overall, this study provides a new measurement method for construction control, health monitoring, intelligent detection, and other fields of cable‐stayed bridges.

A task-driven cerebral angiographic imaging based on CT perfusion.

Current clinical computed tomography arteriography (cCTA) and clinical computed tomography venography (cCTV) images often display restricted cerebrovascular profiles, incomplete brain tissue segmentation, and incomplete artery-vein segmentation. Especially for vessels associated with diseases, capturing their complete profiles proves challenging. In this work, we developed a Task-driven Cerebral Angiographic Imaging (TDCAI) technique using computed tomography perfusion (CTP) images of stroke patients. A evaluation on intracranial hemorrhagic stroke (IHS) and acute ischemic stroke (AIS) cases was performed with CT perfusion imaging. The TDCAI technique processed the CTP images, resulting in supplementary diagnostic images, including CTA, CTV, centerline images of the vessels-of-interest [internal carotid artery (ICA) for AIS patients, Labbé vein for IHS patients], and straightened images of the vessels-of-interest. We conducted a comparison between the obtained CTA/CTV images and the cCTA/cCTV images in terms of overall image quality and visibility of the vessels-of-interest. By constructing a virtual vascular phantom, we extracted its centerline and compared it with the actual centerline to calculate maximum and average deviations. This allowed us to evaluate both the accuracy of the centerline extraction algorithm and its capability to resist the influence of side branches. We assessed whether vascular stenosis and dilatation could be expressed in straightened vessel images, conducting statistical analyses to establish the superiority of TDCAI technique. This study proposes a TDCAI technique to eliminate bone and soft tissue interference, effectively segregate the comprehensive cerebral venous and arterial systems, and extract centerlines and straighten the vessels-of-interest, which would aid doctors in assessing the outflow profiles of vessels after a stroke and seeking imaging biomarkers correlated with clinical outcomes.

Multi-Crop Navigation Line Extraction Based on Improved YOLO-v8 and Threshold-DBSCAN under Complex Agricultural Environments

Field crops are usually planted in rows, and accurate identification and extraction of crop row centerline is the key to realize autonomous navigation and safe operation of agricultural machinery. However, the diversity of crop species and morphology, as well as field noise such as weeds and light, often lead to poor crop detection in complex farming environments. In addition, the curvature of crop rows also poses a challenge to the safety of farm machinery during travel. In this study, a combined multi-crop row centerline extraction algorithm is proposed based on improved YOLOv8 (You Only Look Once-v8) model, threshold DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering, least squares method, and B-spline curves. For the detection of multiple crops, a DCGA-YOLOv8 model is developed by introducing deformable convolution and global attention mechanism (GAM) on the original YOLOv8 model. The introduction of deformable convolution can obtain more fine-grained spatial information and adapt to crops of different sizes and shapes, while the combination of GAM can pay more attention to the important feature areas of crops. The experimental results shown that the F1-score and mAP value of the DCGA-YOLOv8 model for Cabbage, Kohlrabi, and Rice are 96.4%, 97.1%, 95.9% and 98.9%, 99.2%, 99.1%, respectively, which has good generalization and robustness. A threshold-DBSCAN algorithm was proposed to implement clustering for each row of crops. The correct clustering rate for Cabbage, Kohlrabi and Rice reaches 98.9%, 97.9%, and 100%, respectively. And LSM and cubic B-spline curve methods were applied to fit straight and curved crop rows, respectively. In addition, this study constructed a risk optimization function for the wheel model to further improve the safety of agricultural machines operating between crop rows. This indicates that the proposed method can effectively realize the accurate recognition and extraction of navigation lines of different crops in complex farmland environment, and improve the safety and stability of visual navigation and field operation of agricultural machines.

Accurate laser centerline extraction algorithm used for 3D reconstruction of brake caliper surface

Accurate laser centerline extraction algorithm used for 3D reconstruction of brake caliper surface

Automated Coronary Artery Tracking with a Voronoi-Based 3D Centerline Extraction Algorithm.

Coronary artery disease is one of the leading causes of death worldwide, and medical imaging methods such as coronary artery computed tomography are vitally important in its detection. More recently, various computational approaches have been proposed to automatically extract important artery coronary features (e.g., vessel centerlines, cross-sectional areas along vessel branches, etc.) that may ultimately be able to assist with more accurate and timely diagnoses. The current study therefore validated and benchmarked a recently developed automated 3D centerline extraction method for coronary artery centerline tracking using synthetically segmented coronary artery models based on the widely used Rotterdam Coronary Artery Algorithm Evaluation Framework (RCAAEF) training dataset. Based on standard accuracy metrics and the ground truth centerlines of all 32 coronary vessel branches in the RCAAEF training dataset, this 3D divide and conquer Voronoi diagram method performed exceptionally well, achieving an average overlap accuracy (OV) of 99.97%, overlap until first error (OF) of 100%, overlap of the clinically relevant portion of the vessel (OT) of 99.98%, and an average error distance inside the vessels (AI) of only 0.13 mm. Accuracy was also found to be exceptionally for all four coronary artery sub-types, with average OV values of 99.99% for right coronary arteries, 100% for left anterior descending arteries, 99.96% for left circumflex arteries, and 100% for large side-branch vessels. These results validate that the proposed method can be employed to quickly, accurately, and automatically extract 3D centerlines from segmented coronary arteries, and indicate that it is likely worthy of further exploration given the importance of this topic.

Smart warehouse track identification based on Res2Net-YOLACT+HSV

In recent years, the rapid development of the courier industry, in order to solve the problem of the accumulation of goods in courier warehouses, low efficiency of manual sorting and transportation, many courier companies have introduced intelligent unmanned warehouse logistics systems for unmanned transport vehicles or unmanned inspection The study proposes a recognition method based on the combination of Res2Net-YOLACT and HSV to achieve accurate recognition of the route trajectory of unmanned vehicles in the logistics warehouse environment. After the detection of the YOLACT network with improved backbone network, the area of the track is extracted by HSV, and the extracted area is used as the object of secondary processing, i.e., the extracted track is then subjected to centerline extraction to form the actual track. By comparing the recognition effect of this algorithm on the test set under different direction shapes of tracks, the correct rate of Res2Net-YOLACT in the same experimental environment is 97.37%, and the speed of detecting a single image is 30.26 ms, and then the centerline extraction algorithm is ported to the Res2Net-YOLACT network, and the speed of detecting a single image is 37.26 ms. Compared with the 93.28% accuracy of the original YOLACT network, the speed of single image detection is improved by 3.6%. In addition, the centerline extraction algorithm designed using this study is less computationally intensive and less code engineering than the prevailing algorithm, increasing the memory footprint by only 3.2%. To verify the performance of this centerline extraction algorithm, a comparison between this algorithm and the commonly available algorithms was performed, which showed a 1.31% improvement in correctness and an 8.73% improvement in the speed of detecting videos compared to the general algorithm, indicating that the centerline extraction algorithm processed by this study has higher accuracy and real-time performance without significantly consuming more memory. In addition, to test the practicality of the algorithm, the time used by the algorithm to detect the same video on the embedded device jetson nano was counted, and the average frame rate was calculated to be 28FPS and the maximum frame rate was 33FPS, which can be achieved in real transportation applications.

Fast and accurate centerline extraction algorithm for a laser stripe applied for shoe outsole inspection.

Line laser 3D reconstruction technology is widely used in industrial applications. As a key step of this technology, line laser midline extraction directly affects the accuracy of the 3D reconstructed model. In reconstructing the shoe outsole, the traditional algorithm based on the threshold method to determine the laser position may result in a large amount of information loss and miscellaneous point misjudgment owing to the irregularity of the shoe outsole surface, which critically affects the laser imaging quality. To address this problem, an algorithm based on the QQ plot inspection of the laser has been proposed. The QQ plot is a scatter plot, the abscissa is usually the quantile of the standard normal distribution, and the ordinate is the quantile of the data to be tested. If the points on the scatter plot tend to be straight lines, the data to be tested is in a normal distribution. Based on this property, the proposed algorithm aims to check whether the pixels of the image column tend to be normally distributed, rather than using traditional thresholding methods to locate the laser. The objective is to examine whether the image column pixel distribution is normal, instead of using the traditional threshold method to locate the laser. However, the calculation speed of this method is extremely low. To enhance the efficiency of testing the normality of the QQ plot, a quantile-repetition (Q-R) test method is proposed. In this approach, the degree of repetition of quantiles and the position of Q-R values are used to replace the QQ plot based evaluation of the points being on a straight line, and the exact center position is determined by the GGM. The experimental results show that the proposed algorithm can extract more effective points and fewer invalid points of the laser compared to those obtained using the traditional approach, in a rapid, stable, and accurate manner.

A multiview-based automatic method for 3D fracture line extraction

The 3D fracture line is the centerline of the crack region on the surface of the bone that can effectively help orthopedists understand fracture distribution and aid in fracture treatment. However, no effective automatic extraction method exists, and fracture lines still require manual work. This paper develops an efficient automatic method for fracture line extraction from multiple reassembled fragments. In the initial step, we propose a multiview sampling strategy based on spherical uniform viewpoints, which uses off-screen rendering to sample fragments from multiple viewpoints. Markers are generated from the samples and integrated from all viewpoints. Then, we use a new point-based centerline extraction algorithm to merge markers and extract the fracture lines. Finally, we propose a normal-based fracture line mapping method to quickly and reliably map fracture lines to the surface of a standard bone template. Experiments indicate that our method can extract fracture lines from six common complex fragments within 3 s. Through multiview experiments, we select 30 viewpoints so that most template mesh triangles corresponding to the six fragments can be simultaneously observed from at least 2 viewpoints. Additionally, anti-noise experiments show that the proposed method can handle high-noise and cancellous fragments reliably. For complex fragments, our method is still applicable, has strong practicality, and can effectively help doctors save time.

Metal surface 3D imaging based on convolutional autoencoder and light plane calibration

Convolutional autoencoders play an important role in computer vision and image processing. In this paper, an optical stripe extraction method based on convolutional autoencoder is proposed to reconstruct reflective metal surfaces. A six-layer convolutional autoencoder network is constructed to extract the center line of the light stripe. In addition, combined with the center line extraction algorithm and line-structured light calibration algorithm, a good reconstruction effect under reflective conditions is obtained. Experimental results show that the relative error of the 3D imaging method based on the convolutional autoencoder is decreased from 6.233% to 2.183% compared with the traditional line-structured laser 3D imaging method.

Extraction of centerline of large-format laser stripes against ambient light interference

In order to carry out real-time 3D reconstruction of large components in the outdoor environment, this paper proposes a laser stripe centerline extraction algorithm that is resistant to ambient light interference. The median filter is used to preprocess the laser stripe to be extracted, and the Rosenfeld thinning and least square fitting algorithm are organically combined to extract the centerline of the laser stripe. Experimental results show that this method can effectively avoid strong light interference and accurately obtain the centerline of the laser stripe to be extracted.

3D measurement system based on divergent multi-line structured light projection, its accuracy analysis

This paper proposes an effective three-dimensional(3D) measurement system based on divergent multi-line structured light projection, and analyzes the measurement accuracy of the system. In the system, the previous methods have been improved. The accuracy of system calibration and 3D measurement can be improved effectively by changing the calibration board, changing the relative distance between the camera and the laser. Firstly, the camera calibration and the light plane calibration are realized by projecting a circular calibration board and a blank board. Secondly, the Steger centerline extraction algorithm is used to complete the extraction of the centerline of the stripe. Finally, based on the system calibration parameters and the centerline extraction results, the surface 3D measurement of the object is realized, and the camera calibration, light plane calibration and 3D measurement accuracy are analyzed. Experiments show that the calibration method proposed in this paper can obtain higher accuracy than the checkerboard calibration method, which can reduce the camera calibration error by more than 3 times and the light plane calibration error by more than 1 times. In addition, compared with the previous calibration method, the blank board avoids the influence of the calibration pattern on the light stripe. Finally, the relative error of light plane calibration is controlled within 1% by changing the relative distance between the camera and the laser.

DanioSense: Automated High-Throughput Quantification of Zebrafish Larvae Group Movement

The capability to obtain detailed motility information of model organisms is fundamental to reveal their functional and social behavior characteristics. Zebrafish is a powerful vertebrate model organism. Despite recent success in the automatic quantification of adult zebrafish movement, it remains a laborious task for group zebrafish larval tracking due to their similar appearance, frequent occlusions, and highly discontinuous kinematics. This article presents DanioSense (DS), an automatic tracker for group larval zebrafish, to overcome these tracking challenges. The integration of a light convolutional neural network and a centerline extraction algorithm enables the tracker to localize individuals even in occlusion cases where objects' identities are prone to switch. With reliable detections, an adaptive Kalman filter is designed to optimally estimate locomotive parameters, which is also used for object reidentification accomplished by a two-stage data association protocol. Experimental results demonstrated a tracking accuracy of over 97%, median errors of 102 μm, and 8.8° for the position and orientation measurement, and a processing speed of over 30 frames/s with a normal computer configuration. DS provides detailed quantitative data for a large-scale larvae group in nearly real time, highly boosting the efficiency of characterizing individual phenotypes and analyzing social interactions.

An Intestinal Centerline Extraction Algorithm Based on Federated Framework

The intestine is an important organ of the human body, and its internal structure always needs to be observed in clinical applications so as to provide a basis for accurate diagnosis. However, due to the limited intestinal data obtained by a single institution, deep learning cannot effectively train the intestines, and the effect is not satisfied. For this reason, we propose a distributed training method to carry out federated learning to alleviate the situation of patient sample data shortage, not shared and uneven data distribution. And the blockchain is introduced to enhance the interaction between networks, to solve the problem of a single point of failure of the federated learning server. Fully excavate the multiscale features of samples, to construct a fusion enhancement model and intestinal segmentation module for accurate positioning. At the local end, the centerline extraction algorithm is optimized, with the edge as the main and the source as the auxiliary to realize centerline extraction.

Hardware-Oriented Algorithm for High-Speed Laser Centerline Extraction Based on Hessian Matrix

Centerline extraction is the basic and key procedure in line-structured laser 3-D scanners. In this article, we propose a hardware-oriented algorithm for fast and accurate extraction of a laser centerline based on the Hessian matrix. The algorithm is divided into three low-coupling modules that can be processed in parallel—coarse positioning, linewidth estimation, and precise positioning module. In the coarse positioning module, a window slider is used to traverse all image pixels in the raster-scan mode for collecting local image features, based on which the potential region of interest (ROI) containing laser lines is detected. In the linewidth estimation module, the second-order moment features in the detected ROI are calculated to estimate the linewidth according to the local rectangular similarity characteristics of the laser line. In the precise positioning module, the optimal Gaussian template estimated by the linewidth is convolved with the ROI to obtain the Hessian matrix. Based on the Hessian matrix, the normal direction of the laser line is obtained, and the second-order Taylor expansion is performed in that direction to determine the subpixel position of the center point. Finally, non-maximum suppression is used to remove noise points and obtain the most reliable single-pixel centerline. The proposed algorithm is evaluated using thousands of sample images with different materials, exposure times, and laser line shapes, and it presents high robustness for subpixel precision extraction. By implementing the proposed algorithm on a field-programmable gate array (FPGA), a high-speed, high-precision laser centerline extraction system is achieved, which operates at 1350 frames/s for a 1-million-pixel video stream.

The Centerline Extraction Algorithm of Weld Line Structured Light Stripe Based on Pyramid Scene Parsing Network

Based on the good feature learning ability of the pyramid scene parsing network, a method for extracting the centerline of structured light stripes of weld lines based on the pyramid scene parsing network and Steger algorithm is proposed. This method avoids the traditional complex weld image preprocessing technology, and simplifies the operation steps of extracting the centerline of the structured light stripe of the weld image line. In this paper, the pyramid scene parsing network is used to predict the pixels containing weld feature information. Through the pyramid pooling module, the local and global context feature information is fused to supplement the feature information of the weld edge, and then the Steger algorithm is used to extract the weld feature centerline. The results show that the method in this paper can accurately extract the centerline position of the structured light stripe of the weld line under the interference of reflection, and the average value reaches 86.8% on the accuracy evaluation index mean intersection over union, the 18.93 pixels on the weld extraction accuracy index root mean square error, and the average time of extracting the center line of structured light stripe of weld line is <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.188s$ </tex-math></inline-formula> .

Visual inspection of weld surface quality

Welding is an important method for modern material processing. In actual processing, due to the influence of processing accuracy and welding thermal deformation, various defects often appear in the appearance of the weld. At present, visual inspection is mainly used for the appearance inspection of welds. The detection of weld defects mainly depends on the work experience of the staff. Based on the above background, the purpose of this article is to visually inspect the weld surface quality. This article uses visually obtained fringe images of weld contours as information sources to explore a visual-based weld appearance detection algorithm, including the measurement of weld formation dimensions and the detection of weld appearance defects. This algorithm overcomes manual measurements of the misjudgments and omissions caused by eye fatigue and experience differences. It improves the efficiency and accuracy of welding appearance inspection, and meets the needs of automation and intelligence of the entire welding process. In this paper, a subpixel stripe centerline extraction algorithm based on the combination of the Hessian matrix method and the center of gravity method is used; to further improve the accuracy of the extraction of the centerline of the weld seam, this article also performs the work of removing the wrong points and the compensation of the broken seam. Obtain a fringe centerline with better connectivity. Comparing the extraction algorithms of each centerline, the centerline obtained by this method has high accuracy, less time-consuming and high stability. It laid the foundation for the subsequent inspection of weld appearance. Through the training of the model, the accurate classification and recognition of surface defects of tube and plate welds have been achieved. The experimental results show that the improved vision-based welding surface defect recognition and classification proposed in this paper has better performance and accuracy. Up to 96.34%.

A novel centerline extraction algorithm for a laser stripe applied for turbine blade inspection

Line-laser 3D scanning is a novel and promising automation technique for turbine blade inspection. One key task encountered in line-laser 3D scanning is centerline extraction of the laser stripes. In some regions, there is asymmetry of the line profiles caused by high curvature, such as the leading and trailing edges of a blade. According to the asymmetric Gaussian distribution line profile, this paper presents a new algorithm to extract the centerline of a laser stripe based on sub-pixel boundary extraction and medial axis transformation. The candidate boundary points of the laser stripe are extracted by zero crossing. Based on Taylor expansion and interpolation, the gray gradient distribution of the boundary can be estimated accurately. Following this, the accurate boundary points are extracted by extreme values of the gray gradient, and the centerline is calculated by medial axis transformation of the accurate boundary points. The proposed method can effectively reduce errors caused by the asymmetry of line profiles in areas of high curvature. The effectiveness of this method is verified using simulation and measuring experiments. The experiment results show that the extraction error of the asymmetric Gaussian line profile is within ± 0.15 pixels, and the root mean square error in measurement accuracy experiments does not exceed 0.015 mm.

River centerline extraction using the multiple direction integration algorithm for mixed and pure water pixels

The river centerline is a basic hydrological characteristic. Most prior studies have used remote sensing data to extract the river centerline from the open water region in a pure water pixel region. Extracting this type of river is relatively easy. However, extracting the centerline of a micro-river, which is mainly composed of mixed water pixels, is challenging. This paper presents a novel method, called the Multiple Direction Integration Algorithm (MDIA), to extract the river centerline using an image-enhancing method combined with river morphology. MDIA can be applied to regions mainly composed of pure water pixels, as well as to regions consisting of mixed water pixels in the index image. The method first calculates the normalized difference vegetation index (NDVI) and enhances the river linear structure using a Hessian matrix. Second, a small window is constructed as a circular structural element. In the window region, the local threshold is automatically obtained using water-oriented clustering segmentation and prior river knowledge to judge the pixel type. After completing the river centerline extraction in the current window, the next detecting window is generated to continue judgment. The structural element automatically executes river centerline judgment until the entire river centerline is extracted. The Landsat 8 images of six regions with different geomorphologies were chosen to analyze the method’s performance. The test sites include high mountain region, low mountain region, plains region with farmland and a residential region. The experimental results show that the optimal threshold of the processing results ranged from 0.2 to 0.3. In this range, the user’s accuracy is 0.813 to 0.997, and the producer’s accuracy is 0.981 to 1. The MDIA effectively and correctly extracts the river network in mixed-pixel regions. The presented method provides an effective algorithm for river centerline extraction that can be used to expand and update river datasets and provide reliable river centerline data for relevant hydrology studies.

3-D Human Sperm Flagellum Tracing in Low SNR Fluorescence Images.

Tracing tubular structures from biomedical images is important for a wide range of applications. Particularly, the spermatozoon is an essential cell whose flagella have a tubular form. Its main function is to fertilize the egg, and the flagellum is fundamental to achieve this task which depends importantly on the dynamics of intracellular calcium ([Ca2+]i). Measuring [Ca2+]i along the flagellum in 3-D is not a simple matter since it requires: 1) sophisticated fluorescence imaging techniques dealing with low intensity and signal to noise ratio (SNR) and 2) tracing the flagellum's centerline. Most of the algorithms proposed to trace tubular structures have been developed for multi-branch structures not being adequate for single tubular structures with low SNR. Taking into account the prior knowledge that the flagellum is constituted by a single tubular structure, we propose an automatic method to trace and track multiple single tubular structures from 3-D images. First, an algorithm based on one-class classification allows enhancement of the flagellum. This enhanced 3-D image permits guiding an iterative centerline algorithm toward the flagellum's centerline. Each sperm is assigned an ID to keep track of it in 3-D . Our algorithm was quantitatively evaluated using a ground truth 564 semi-manual traces (six 3-D image stacks) comparing them to those obtained from state-of-the-art tubular structure centerline extraction algorithms. The qualitative and quantitative results show that our algorithm is extracting similar traces as compared with ground truth, and it is more robust and accurate to trace the flagellum's centerline than multi-branch algorithms.