Euclidean Distance Transform Research Articles

Panoptic Segmentation with Convex Object Representation

Abstract The accurate representation of objects holds pivotal significance in the realm of panoptic segmentation. Presently, prevalent object representation methodologies, including box-based, keypoint-based and query-based techniques, encounter a challenge known as the ‘representation confusion’ issue in specific scenarios, often resulting in the mislabeling of instances. In response, this paper introduces Convex Object Representation (COR), a straightforward yet highly effective approach to address this problem. COR leverages a CNN-based Euclidean Distance Transform to convert the target instance into a convex heatmap. Simultaneously, it offers a parallel embedding method for encoding the object. Subsequently, COR characterizes objects based on the distinctive embedding vectors of their convex vertices. This paper seamlessly integrates COR into a state-of-the-art query-based panoptic segmentation framework. Experimental findings validate that COR successfully mitigates the representation confusion predicament, enhancing segmentation accuracy. The COR-augmented methods exhibit notable improvements of +1.3 and +0.7 points in PQ on the Cityscapes validation and MS COCO panoptic 2017 validation datasets, respectively.

Feature information extraction method for narrow gap U-type groove based on laser vision

The precise measurement of weld groove position and geometric dimensions serves as the foundation for predicting and planning welding strategies and process parameters. The steep sidewalls of narrow gap U-type grooves pose challenges for sensing and detection during the welding process. This paper presents a feature information extraction method for narrow gap U-type grooves based on laser vision technology. Firstly, the image acquisition system is calibrated using the incident plane of the laser beam. Subsequently, a digital image processing technique is employed to enhance the collected images, combining a 3D spatial domain enhancement algorithm with a laser stripe image refinement method based on Euclidean distance transformation. Finally, the transverse and vertical positions of the groove, as well as its geometric features such as width, depth, assembly gap, and misalignment, are computed and measured using feature point positioning. The proposed algorithm is experimentally verified for its accuracy, robustness, and real-time performance. The results indicate that the method effectively suppresses unevenness in laser stripe brightness and mitigates random noise in the captured images. The average measurement errors for transverse position, vertical position, width, depth, assembly gap, and misalignment of the groove are 0.0857 mm, 0.0987 mm, 0.1664 mm, 0.2216 mm, 0.0935 mm, and 0.0997 mm, respectively. Furthermore, the average image processing time per frame is 58 ms, meeting the requirements of pre-weld detection in arc welding.

Inspection and classification of wheat quality using image processing

Wheat plays an important role in our daily life and industrial production. Several computer vision approaches have been proposed for classifying wheat quality, but there were some methods focusing on the problem of cohesive wheats while image processing. In this paper, we designed a single kernel guide groove to separate the cohesive wheats, which could simplify the algorithm of image processing and improve the accuracy rate of classification. For the method followed while recording the data, the image information must be converted into digital information, and the results are provided using appropriate image processing algorithms. Image preprocessing steps such as binarization, image enhancement, image segmentation, and morphological processing were used to reduce noise. For image segmentation, we proposed the following new segmentation methods: (1) extracting wheat region by converting image to H channel and (2) watershed algorithm based on Euclidean distance transformation. For the classification model, 22 features of 7 different qualities of wheat were inputted in the Back Propagation (BP) neural network and Support Vector Machine (SVM) model, and the overall correct classification rates were determined to be 91% and 97% for SVM and BP neural network, respectively. The BP neural network was more suitable for wheat appearance quality detection.

Analysis of density-dependent bead and cell structure of expanded polypropylene bead foams from X-ray computed tomography of different resolution

Closed-cell bead foams with their hierarchical geometrical structure are a challenge for statistical reconstruction and finite element modelling. For the purpose of providing the fundamental micro- and meso-structural descriptors - wall thickness, cell as well as bead volume and sphericity - of expanded polypropylene bead foams of different density, 3D-images from X-ray computed tomography are analyzed. A detailed description of development and application of an image analysis methodology for the determination of feature distributions from CT-scans of different level of detail is provided. The methods are based on off-the-shelf algorithms provided by the open-source package distribution FIJI. It should be highlighted, that beside essential methods such as thresholding, euclidean distance and watershed transformation here the Trainable WEKA segmentation is applied for separating material phases in the images. Although the methods elaborated are generally very case sensitive, the reader benefits from the validation strategies applied, so that development of individual methods into the direction of reliability, repeatability and automation is supported.

Prediction method of dust pollutant diffusion range in building demolition based on Euclidean distance transformation

Aiming at the problem of low prediction accuracy in the prediction of pollutant diffusion range, a prediction method of dust pollutant diffusion range in building demolition based on Euclidean distance transformation is proposed. The Euclidean distance field outside the collection object is established by Euclidean distance transformation, and the parameters of dust pollutants in building demolition are extracted. According to the atmospheric stability and the position of the boundary layer, the Gaussian model is used to simulate the diffusion trend of fugitive dust pollutants. The prediction model is then established with the help of cellular automata model. The parameters are input into the prediction model, and the output result is the prediction result to realise the prediction. The results show that the prediction accuracy of the proposed method is high.

GPU-Accelerated Incremental Euclidean Distance Transform for Online Motion Planning of Mobile Robots

In this letter, we present a volumetric mapping system that effectively calculates Occupancy Grid Maps (OGMs) and Euclidean Distance Transforms (EDTs) with parallel computing. Unlike these mappers for high-precision structural reconstruction, our system incrementally constructs global EDT and outputs high-frequency local distance information for online robot motion planning. The proposed system receives multiple types of sensor inputs and constructs OGM without down-sampling. Using GPU programming techniques, the system quickly computes EDT in parallel within local volume. The new observation is continuously integrated into the global EDT using the parallel wavefront algorithm while preserving the historical observations. Experiments with datasets have shown that our proposed approach outperforms existing state-of-the-art robot mapping systems and is particularly suitable for mapping unexplored areas. In its actual implementations on aerial and ground vehicles, the proposed system achieves real-time performance with limited onboard computational resources.

Trajectory tracking for quadrotors: An optimization‐based planning followed by controlling approach

AbstractWe present an optimization‐based reference trajectory tracking method for quadrotor robots for slow‐speed maneuvers. The proposed method uses planning followed by the controlling paradigm. The basic concept of the proposed method is an analogy with linear quadratic Gaussian in which nonlinear model predictive control (NMPC) is employed for predicting optimal control policy in each iteration. Multiple‐shooting is suggested over direct‐collocation for imposing constraints when modeling the NMPC. Incremental Euclidean distance transformation map is constructed for obtaining the closest free distances relative to the predicted trajectory; these distances are considered obstacle constraints. The reference trajectory is generated ensuring dynamic feasibility. The objective is to minimize the error between the quadrotor's current pose and the desired reference trajectory pose in each iteration. Finally, we compared the proposed method with two other approaches and showed that the proposed method outperforms the said approaches in terms of reaching the goal without any collision. Additionally, we published a new data set that can be used for evaluating the performance of trajectory tracking algorithms.

Prediction method of dust pollutant diffusion range in building demolition based on Euclidean distance transformation

Prediction method of dust pollutant diffusion range in building demolition based on Euclidean distance transformation

SUBDIVISION SURFACE MID-SURFACE RECONSTRUCTION OF TOPOLOGY OPTIMIZATION RESULTS AND THIN-WALLED SHAPES USING SURFACE SKELETONS

AbstractPolygon meshes and particularly triangulated meshes can be used to describe the shape of different types of geometry such as bicycles, bridges, or runways. In engineering, such polygon meshes can occur as finite element meshes, resulting from topology optimization or laser scanning. This article presents an automated parameterization of polygon meshes into a parametric representation using subdivision surfaces, especially in topology optimization. Therefore, we perform surface skeletonization on a volumetric grid supported by the Euclidian distance transformation and topology preserving and shape-preserving criterion. Based on that surface skeleton, an automated conversation into a Subdivision Surface Control grid is established. The final mid-surface-like parametrization is quite flexible and can be changed by variating the control gird or the local thickness.

Cortical Thickness Estimation: A Comparison of FreeSurfer and Three Voxel-Based Methods in a Test-Retest Analysis and a Clinical Application.

The cortical thickness has been used as a biomarker to assess different cerebral conditions and to detect alterations in the cortical mantle. In this work, we compare methods from the FreeSurfer software, the Computational Anatomy Toolbox (CAT12), a Laplacian approach and a new method here proposed, based on the Euclidean Distance Transform (EDT), and its corresponding computational phantom designed to validate the calculation algorithm. At region of interest (ROI) level, within- and inter-method comparisons were carried out with a test-retest analysis, in a subset comprising 21 healthy subjects taken from the Multi-Modal MRI Reproducibility Resource (MMRR) dataset. From the Minimal Interval Resonance Imaging in Alzheimer's Disease (MIRIAD) data, classification methods were compared in their performance to detect cortical thickness differences between 23 healthy controls (HC) and 45 subjects with Alzheimer's disease (AD). The validation of the proposed EDT-based method showed a more accurate and precise distance measurement as voxel resolution increased. For the within-method comparisons, mean test-retest measures (percentages differences/intraclass correlation/Pearson correlation) were similar for FreeSurfer (1.80%/0.90/0.95), CAT12 (1.91%/0.83/0.91), Laplacian (1.27%/0.89/0.95) and EDT (2.20%/0.88/0.94). Inter-method correlations showed moderate to strong values (R > 0.77) and, in the AD comparison study, all methods were able to detect cortical alterations between groups. Surface- and voxel-based methods have advantages and drawbacks regarding computational demands and measurement precision, while thickness definition was mainly associated to the cortical thickness absolute differences among methods. However, for each method, measurements were reliable, followed similar trends along the cortex and allowed detection of cortical atrophies between HC and patients with AD.

EDT Method for Multiple Labelled Objects Subject to Tied Distances

The success of new scientific areas can be assessed by their potential for contributing to new theoretical approaches aligned with real-world applications. The Euclidean distance transform (EDT) has fared well in both cases, providing a sound theoretical basis for a number of applications, such as median axis transform, fractal analysis, skeletonization, and Voronoi diagrams. Despite its wide applicability, the discrete form of the EDT includes interesting properties that have not yet been fully exploited in the literature. In this paper, we are particularly interested in the properties of 1) working with multiple objects/labels; and 2) identifying and counting equidistant pixels/voxels from certain points of interest. In some domains (such as dataset classification, texture, and complexity analysis), the result of applying the EDT transform with different objects, and their respective tied distances, may compromise the performance. In this sense, we propose an efficient modification in the method presented in [ 1 ], which leads to a novel approach for computing the distance transform in a space with multiple objects, and for counting equidistant pixels/voxels.

Mechanisms of Confining Pressure Dependence of Resistivity Index for Tight Sandstones by Digital Core Analysis

SummaryResistivity measurements are a major input into hydrocarbon reserve estimation and are usually described by Archie’s laws. In this study, we use digital rock physics to analyze the mechanisms of non-Archie and Archie behavior of formation factor (FF) and resistivity index (RI) of low-porosity Fontainebleau (FB) sandstone for ambient conditions and under high confining pressure, respectively.FB sandstone was imaged by micro-X-ray computed tomography (micro-CT) at a resolution of 1 µm. Subresolution details of the grain contact width distribution along with their length were extracted from a set of scanning electron microscope (SEM) images. The nanoscale aperture of grain contacts, which is below tomogram resolution, is accounted for in micro-CT-based numerical calculations by assigning effective porosity and conductivity to individual voxels of the extracted grain contact network. A porosity reduction of grain contacts and open pore space as a function of applied confining pressure is introduced, capturing the pressure dependence. The concept was implemented by grain contact labeling, introducing an additional phase derived from a Euclidean distance transform (EDT). Subvoxel stress-strain effects were incorporated by attributing all compressibility effects to the pore space (open pore space and grain contacts), treating the solid phase as perfectly rigid. Voxel-scale input conductivities are assigned using Archie’s law followed by solving the Laplace equation for sample-scale rock resistivity and RI directly on the segmented image using the finite element method.For the numerical modeling of the FF and RI of low-porosity FB sandstone as a function of confining pressure, which depends on subresolution features, a set of hypotheses were tested. These are based on two segmentation scenarios incorporating the measured contact aperture distribution from SEM analysis—a homogeneous aperture-based segmentation by assuming all grain contacts as an average constant value and a heterogeneous aperture-based segmentation assigning two groups of grain contact apertures. The segmentation scenarios enable homogeneous and heterogeneous morphological change of grain contacts due to confining pressure effects. Furthermore, partial saturation of grain contacts is considered. In all cases, strong water-wetness was assumed, and discretization effects were analyzed carefully.The numerical results highlight the relative contribution of each of two conductive components of FB sandstone (open pores vs. grain contacts) over the full range of partial saturations. Of importance is the connectivity of the system, with discretization effects having a significant effect on FF, but a small effect on the RI. Grain contacts and confining pressure are found to have a significant impact on RI behavior of low-porosity FB sandstone. Both the grain contact network with homogeneous aperture and the heterogeneous grain contact network are able to describe experimental observations. However, it is not sufficient to assume a homogeneous change in contact area, and an inhomogeneous deformation of grain contact zones is required to match the experiment.

Semantic Recognition of Human-Object Interactions via Gaussian-Based Elliptical Modeling and Pixel-Level Labeling

Human-Object Interaction (HOI) recognition, due to its significance in many computer vision-based applications, requires in-depth and meaningful details from image sequences. Incorporating semantics in scene understanding has led to a deep understanding of human-centric actions. Therefore, in this research work, we propose a semantic HOI recognition system based on multi-vision sensors. In the proposed system, the de-noised RGB and depth images, via Bilateral Filtering (BLF), are segmented into multiple clusters using a Simple Linear Iterative Clustering (SLIC) algorithm. The skeleton is then extracted from segmented RGB and depth images via Euclidean Distance Transform (EDT). Human joints, extracted from the skeleton, provide the annotations for accurate pixel-level labeling. An elliptical human model is then generated via a Gaussian Mixture Model (GMM). A Conditional Random Field (CRF) model is trained to allocate a specific label to each pixel of different human body parts and an interaction object. Two semantic feature types that are extracted from each labeled body part of the human and labelled objects are: Fiducial points and 3D point cloud. Features descriptors are quantized using Fisher's Linear Discriminant Analysis (FLDA) and classified using K-ary Tree Hashing (KATH). In experimentation phase the recognition accuracy achieved with the Sports dataset is 92.88%, with the Sun Yat-Sen University (SYSU) 3D HOI dataset is 93.5% and with the Nanyang Technological University (NTU) RGB+D dataset it is 94.16%. The proposed system is validated via extensive experimentation and should be applicable to many computer-vision based applications such as healthcare monitoring, security systems and assisted living etc.

Retinal image blood vessel extraction and quantification with Euclidean distance transform approach

Image processing applications remarkably contributes to modern ophthalmology. This technology is designed to analyse the characteristics of the human eye microvasculature images. The retinal microvasculature is an excellent non-invasive screening window for the assessment of systemic diseases such as diabetes, hypertension, and stroke. Retinal microvasculature character such as widening vessel diameter is recognised as an analysable feature for stroke or transient ischemic attack for predicting the progression of this pathology. Thus, in this study, a computer-assisted method has been developed for this task applying the Euclidean distance transform (EDT) technique. This newly developed algorithm computes the Euclidean distance of the remaining white pixels on the area of interest. Central Light Reflex Image Set (CLRIS) and Vascular Disease Image Set (VDIS) of Retinal Vessel Image set for Estimation of Width database were used for the performance evaluation of the proposed algorithm that showed 98.1 and 97.7% accurate result for both CLRIS and VDIS, respectively. The significantly high accuracy in this newly developed vessel diameter quantification algorithm indicates excellent potential for further development, evaluation, validation, and integration into ophthalmic diagnostic instruments.

Parallel Raster Scan for Euclidean Distance Transform

Distance transform (DT) and Voronoi diagrams (VDs) have found many applications in image analysis. Euclidean distance transform (EDT) can generate forms that do not vary with the rotation, because it is radially symmetrical, which is a desirable characteristic in distance transform applications. Recently, parallel architectures have been very accessible and, particularly, GPU-based architectures are very promising due to their high performance, low power consumption and affordable prices. In this paper, a new parallel algorithm is proposed for the computation of a Euclidean distance map and Voronoi diagram of a binary image that mixes CUDA multi-thread parallel image processing with a raster propagation of distance information over small fragments of the image. The basic idea is to exploit the throughput and the latency in each level of memory in the NVIDIA GPU; the image is set in the global memory, and can be accessed via texture memory, and we divide the problem into blocks of threads. For each block we copy a portion of the image and each thread applies a raster scan-based algorithm to a tile of m×m pixels. Experiment results exhibit that our proposed GPU algorithm can improve the efficiency of the Euclidean distance transform in most cases, obtaining speedup factors that even reach 3.193.

Station number assignment to abdominal lymph node for assisting gastric cancer surgery

ABSTRACT This paper presents a method of the station number assignment to the abdominal lymph node for assisting gastric cancer surgery. In gastric cancer surgery, surgeons need to dissect not only the stomach but also surrounding lymph nodes. Therefore, it is important to understand the area where the lymph nodes exist prior to surgery. The proposed method assigns station numbers to each lymph node extracted from a CT volume. Since arteries are in close relation to the station numbers around the stomach, we utilise the arteries’ information for assigning the station numbers. The vessel dominant map is introduced to analyse the spatial relationships between the lymph nodes and the surrounding arteries. This map is computed based on three-dimensional Euclidean distance transformation. The proposed method classifies the lymph nodes into eight groups, which are defined by the station numbers, using the vessel dominant map. We evaluated the proposed method using 10 data sets that include 77 lymph nodes. Lymph node extraction and blood vessel labelling were performed manually on these volumes. The proposed method could label lymph node station numbers with an accuracy of 76.6%. This result showed that the proposed method was useful for assisting gastric cancer surgery.

Predicting dark matter halo formation in N-body simulations with deep regression networks

ABSTRACT Dark matter haloes play a fundamental role in cosmological structure formation. The most common approach to model their assembly mechanisms is through N-body simulations. In this work, we present an innovative pathway to predict dark matter halo formation from the initial density field using a Deep Learning algorithm. We implement and train a Deep Convolutional Neural Network to solve the task of retrieving Lagrangian patches from which dark matter haloes will condense. The volumetric multilabel classification task is turned into a regression problem by means of the Euclidean distance transformation. The network is complemented by an adaptive version of the watershed algorithm to form the entire protohalo identification pipeline. We show that splitting the segmentation problem into two distinct subtasks allows for training smaller and faster networks, while the predictive power of the pipeline remains the same. The model is trained on synthetic data derived from a single full N-body simulation and achieves deviations of ∼10 per cent when reconstructing the dark matter halo mass function at z = 0. This approach represents a promising framework for learning highly non-linear relations in the primordial density field. As a practical application, our method can be used to produce mock dark matter halo catalogues directly from the initial conditions of N-body simulations.

Object extraction via deep learning-based marker-free tracking framework of surgical instruments for laparoscope-holder robots.

The surgical instrument tracking framework, especially the marker-free surgical instrument tracking framework, is the key to visual servoing which is applied to achieve active control for laparoscope-holder robots. This paper presented a marker-free surgical instrument tracking framework based on object extraction via deep learning (DL). The surgical instrument joint was defined as the tracking point. Using DL, a segmentation model was trained to extract the end-effector and shaft portions of the surgical instrument in real time. The extracted object was transformed into a distance image by Euclidean Distance Transformation. Next, the points with the maximal pixel value in the two portions were defined as their central points, respectively, and the intersection point of the line connecting the two central points and the plane connecting the two portions was determined as the tracking point. Finally, the object could be fast extracted using the masking method, and the tracking point was fast located frame-by-frame in a laparoscopic video to achieve tracking of surgical instrument. The proposed object extraction via a DL-based marker-free tracking framework was compared with a marker-free tracking-by-detection framework via DL. Using seven in vivo laparoscopic videos for experiments, the mean tracking success rate was 100%. The mean tracking accuracy was (3.9 ± 2.4, 4.0 ± 2.5) pixels measured in u and v coordinates of a frame, and the mean tracking speed was 15 fps. Compared to the reported mean tracking accuracy of a marker-free tracking-by-detection framework via DL, the mean tracking accuracy of our proposed tracking framework was improved by 37% and 23%, respectively. Accurate and fast tracking of marker-free surgical instruments could be achieved in in vivo laparoscopic videos by using our proposed object extraction via DL-based marker-free tracking framework. This work provided important guiding significance for the application of laparoscope-holder robots in laparoscopic surgeries.

Feature Based Face Recognition using Machine Learning Techniques’

Human Face has Numerous unique Features to Distinguish between each other. Face can Identified by distinguishing between face and non-face followed by Identification. Traditionally face recognition uses distinct features Comparison to Identify the Faces which is Complex for larger databases and ambiguous in many scenarios. To improve the accuracy and Scalability Proposed method uses machine learning based Haar Cascade technique for face detection and convolutional neural network is used for feature extraction followed by classification using Euclidean distance and cosine transformation to recognize the face. The results demonstrate the work is performed well in recognizing the face efficiently with different variations.

Semiautomatic Road Extraction Framework Based on Shape Features and LS-SVM from High-Resolution Images

Road network delineation from remote sensing imagery is useful in many applications related to humans in present days as it provides meaningful information which is very helpful in the field of transportation and other support systems of human civilization. However, the road network extraction from remote sensing images is a challenging problem because of various properties of road network altered by different factors. In this paper, a supervised multistage framework based on least square support vector machine (LS-SVM), mathematical morphology and road shape features is proposed to extract road networks from remote sensing images. In the first stage, after denoising the image using the nonlinear filter, the image is segmented using LS-SVM into two classes’ road or non-road but some non-road components like building structures, shadows of trees and other roads like objects are also classified as the road network. In the second stage, the morphology and primary shape features are used to remove the non-road objects. Finally, road centerlines are extracted using a method based on Euclidean distance transformation. The experimental results reflect the accuracy of the proposed work as compared to other road extraction methods.