Search

Uveitis is one of the leading causes of severe vision loss that can lead to blindness worldwide. Clinical records show that early and accurate detection of vitreous inflammation can potentially reduce the blindness rate. In this paper, a novel framework is proposed for automatic quantification of the vitreous on optical coherence tomography (OCT) with particular application for use in the grading of vitreous inflammation. The proposed pipeline consists of two stages, vitreous region segmentation followed by a neural network classifier. In the first stage, the vitreous region is automatically segmented using a U-net convolutional neural network (U-CNN). For the input of U-CNN, we utilized three novel image descriptors to account for the visual appearance similarity of the vitreous region and other tissues. Namely, we developed an adaptive appearance-based approach that utilizes a prior shape information, which consisted of a labeled dataset of the manually segmented images. This image descriptor is adaptively updated during segmentation and is integrated with the original greyscale image and a distance map image descriptor to construct an input fused image for the U-net segmentation stage. In the second stage, a fully connected neural network (FCNN) is proposed as a classifier to assess the vitreous inflammation severity. To achieve this task, a novel discriminatory feature of the segmented vitreous region is extracted. Namely, the signal intensities of the vitreous are represented by a cumulative distribution function (CDF). The constructed CDFs are then used to train and test the FCNN classifier for grading (grade from 0 to 3). The performance of the proposed pipeline is evaluated on a dataset of 200 OCT images. Our segmentation approach documented a higher performance than related methods, as evidenced by the Dice coefficient of 0.988 ± 0.01 and Hausdorff distance of 0.0003 mm ± 0.001 mm. On the other hand, the FCNN classification is evidenced by its average accuracy of 86%, which supports the benefits of the proposed pipeline as an aid for early and objective diagnosis of uvea inflammation.

Cell nuclear detection on digital histopathology images plays an important role on computer-assisted cancer diagnostics. However, lack of manual annotations and variability of cells bring great challenges to fully-supervised learning. Therefore, we propose a Robust Self-Trained Network (RSTN) for cell detection. The backbone is an encoder-decoder trained by distance maps (DMs) generated from dot annotations of nuclei. To save manual efforts, RSTN is designed to involve reliable predicted DMs in optimization and detect cell centers for unknown images automatically. RSTN gains robustness by regularizing the network by dynamic graphs of DM patches. It exploits underlying graph structures and recognizes complex spatial patterns to locate cells of various shapes and colors. Experimental results show that it outperforms several classic and advanced models on both simulated fluorescence microscope images and real pathology slides for cell detection.

The automatic segmentation of intervertebral discs from medical images is an important task in an intelligent clinical system. In this study, a deep learning model for automatic intervertebral disc segmentation based on MultiResUNet is presented. The MultiResUNet easily segments all intervertebral discs. However, when only certain specific intervertebral discs need to be segmented, problems such as segmentation errors, misalignment, and noise occur. In order to solve these problems, the connected-component labeling is used to relabel the training data so that each intervertebral disc has a different label. A local positioning operation method based on distance transform is proposed to segment the lower two intervertebral discs first, and then train the upper intervertebral disc to segment using the distance feature. The results show that the segmentation errors and misalignments of intervertebral discs are greatly reduced, and the accuracy of correct segmentation is increased to about 94%.

This paper proposed a novel river detection algorithm for SAR images with an improved version of OTSU algorithm. The width of the river can also be estimated using the detected results. The gray threshold, which is critical in the detection of the river, can be obtained adaptively by the proposed improved OTSU algorithm. Thus, it is more accurate than the conventional OTSU algorithm. In addition, DEM is employed in the river region extraction process to mitigate the effect of mountain shadows. The target refinement, the skeleton extraction and the endpoint removal algorithms are all devised to acquire the river centerline image. And the distance transformation algorithm is applied to generate the river distance map. Henceforth, one can estimate the river width by combining the previously acquired images. In order to create a more constant river centerline image, we employed a bridge removal procedure before the river skeletonization. The data used for the evaluation is from Sentinel 1A single-polarization data of the Yangtze River near Nanjing City, Jiangsu Province. The experiment results have shown that the proposed algorithm is more effective in the river extraction.

Brain structure segmentation on magnetic resonance (MR) images is important for various clinical applications. It has been automatically performed by using fully convolutional networks. However, it suffers from the class imbalance problem. To address this problem, we investigated how loss weighting strategies work for brain structure segmentation tasks with different class imbalance situations on MR images. In this study, we adopted segmentation tasks of the cerebrum, cerebellum, brainstem, and blood vessels from MR cisternography and angiography images as the target segmentation tasks. We used a U-net architecture with cross-entropy and Dice loss functions as a baseline and evaluated the effect of the following loss weighting strategies: inverse frequency weighting, median inverse frequency weighting, focal weighting, distance map-based weighting, and distance penalty term-based weighting. In the experiments, the Dice loss function with focal weighting showed the best performance and had a high average Dice score of 92.8% in the binary-class segmentation tasks, while the cross-entropy loss functions with distance map-based weighting achieved the Dice score of up to 93.1% in the multi-class segmentation tasks. The results suggested that the distance map-based and the focal weightings could boost the performance of cross-entropy and Dice loss functions in class imbalanced segmentation tasks, respectively.

The virtually error-free segmentation and tracking of densely packed cells and cell nuclei is still a challenging task. Especially in low-resolution and low signal-to-noise-ratio microscopy images erroneously merged and missing cells are common segmentation errors making the subsequent cell tracking even more difficult. In 2020, we successfully participated as team KIT-Sch-GE (1) in the 5th edition of the ISBI Cell Tracking Challenge. With our deep learning-based distance map regression segmentation and our graph-based cell tracking, we achieved multiple top 3 rankings on the diverse data sets. In this manuscript, we show how our approach can be further improved by using another optimizer and by fine-tuning training data augmentation parameters, learning rate schedules, and the training data representation. The fine-tuned segmentation in combination with an improved tracking enabled to further improve our performance in the 6th edition of the Cell Tracking Challenge 2021 as team KIT-Sch-GE (2).

DeepMind presented remarkably accurate protein structure predictions at the CASP14 conference. We explored network architectures incorporating related ideas and obtained the best performance with a 3-track network in which information at the 1D sequence level, the 2D distance map level, and the 3D coordinate level is successively transformed and integrated. The 3-track network produces structure predictions with accuracies approaching those of DeepMind in CASP14, enables rapid solution of challenging X-ray crystallography and cryo-EM structure modeling problems, and provides insights into the functions of proteins of currently unknown structure. The network also enables rapid generation of accurate models of protein-protein complexes from sequence information alone, short circuiting traditional approaches which require modeling of individual subunits followed by docking. We make the method available to the scientific community to speed biological research.One-Sentence SummaryAccurate protein structure modeling enables rapid solution of structure determination problems and provides insights into biological function.

A set of cameras with fisheye lenses have been used to capture a wide field of view. The traditional scan-line stereo algorithms based on epipolar geometry are directly inapplicable to this non-pinhole camera setup due to optical characteristics of fisheye lenses; hence, existing complete 360° RGB-D imaging systems have rarely achieved realtime performance yet. In this paper, we introduce an efficient sphere-sweeping stereo that can run directly on multiview fisheye images without requiring additional spherical rectification. Our main contributions are: First, we introduce an adaptive spherical matching method that accounts for each input fisheye camera’s resolving power concerning spherical distortion. Second, we propose a fast inter-scale bilateral cost volume filtering method that refines distance in noisy and textureless regions with optimal complexity of O(n). It enables real-time dense distance estimation while preserving edges. Lastly, the fisheye color and distance images are seamlessly combined into a complete 360° RGB-D image via fast inpainting of the dense distance map. We demonstrate an embedded 360° RGB-D imaging prototype composed of a mobile GPU and four fisheye cameras. Our prototype is capable of capturing complete 360° RGB-D videos with a resolution of two megapixels at 29 fps. Results demonstrate that our real-time method outperforms traditional omnidirectional stereo and learning-based omnidirectional stereo in terms of accuracy and performance.

This work proposes a multi-task fully convolutional architecture for tree species mapping in dense forests from sparse and scarce polygon-level annotations using hyperspectral UAV-borne data. Our model implements a partial loss function that enables dense tree semantic labeling outcomes from non-dense training samples, and a distance regression complementary task that enforces tree crown boundary constraints and substantially improves the model performance. Our multi-task architecture uses a shared backbone network that learns common representations for both tasks and two task-specific decoders, one for the semantic segmentation output and one for the distance map regression. We report that introducing the complementary task boosts the semantic segmentation performance compared to the single-task counterpart in up to 11% reaching an average user's accuracy of 88.63% and an average producer's accuracy of 88.59%, achieving state-of-art performance for tree species classification in tropical forests., Full version of preprint accepted at ISPRS Journal of Photogrammetry and Remote Sensing

Multiview shape-from-shading (SfS) has achieved high-detail geometry, but its computation is expensive for solving a multiview registration and an ill-posed inverse rendering problem. Therefore, it has been mainly used for offline methods. Volumetric fusion enables real-time scanning using a conventional RGB-D camera, but its geometry resolution has been limited by the grid resolution of the volumetric distance field and depth registration errors. In this paper, we propose a real-time scanning method that can acquire high-detail geometry by bridging volumetric fusion and multiview SfS in two steps. First, we pro-pose the first real-time acquisition of photometric normals stored in texture space to achieve high-detail geometry. We also introduce geometry-aware texture mapping, which progressively refines geometric registration between the texture space and the volumetric distance field by means of normal texture, achieving real-time multiview SfS. We demonstrate our scanning of high-detail geometry using an RGB-D cam-era at ∼20 fps. Results verify that the geometry quality of our method is strongly competitive with that of offline multi-view SfS methods.

While convolutional neural networks (CNNs) have become the first choice for the medical image segmentation, they still lack the critical ingredient of incorporating priors, such as smoothness and boundary shapes. The authors tackle the limitation by developing a novel prior that is boundary-aware in two ways: promoting smoothness without blurring object boundaries and punishing prediction errors according to boundary shapes. They bring the boundary-aware property into effect by weighting the prediction gradients and errors with the distance map. Their prior differs from previous approaches that either over-smooth boundaries or tend to produce rough boundaries. They evaluate their prior alongside the cross-entropy (CE) on a cardiac MRI dataset. Compared to CE alone, their prior improves the Dice score by 1.5% and Hausdorff distance by 53%. It also yielded a faster and more stable learning process.

The watershed analysis has been widely used to segment contacting particles in volumetric images obtained from X-ray Computed Tomography (X-ray CT). However, the watershed analysis cannot discriminate between contacts of two particles and “necks” of peanut-shaped particles. Then, peanut-shaped particles are mistakenly oversegmented as two particles. To addresses this issue, this study developed an improved watershed analysis, which defines a dimensionless parameter α to discriminate contacts and necks. An image morphological reconstruction technique was developed to modify the distance map based on the parameter α. The watershed transform was performed on a modified distance map to overcome oversegmentation. The accuracy and effectiveness of the proposed technique were validated by six sand specimens. The influence of oversegmentation on particle size and shape characterizations were also demonstrated. The results showed that the improved 3D watershed analysis can effectively segment 3D particle contacts while preserving the continuity of necks for different granular soils.

Cardiac image segmentation is a critical process for generating personalized models of the heart and for quantifying cardiac performance parameters. Several convolutional neural network (CNN) architectures have been proposed to segment the heart chambers from cardiac cine MR images. Here we propose a multi-task learning (MTL)-based regularization framework for cardiac MR image segmentation. The network is trained to perform the main task of semantic segmentation, along with a simultaneous, auxiliary task of pixel-wise distance map regression. The proposed distance map regularizer is a decoder network added to the bottleneck layer of an existing CNN architecture, facilitating the network to learn robust global features. The regularizer block is removed after training, so that the original number of network parameters does not change. We show that the proposed regularization method improves both binary and multi-class segmentation performance over the corresponding state-of-the-art CNN architectures on two publicly available cardiac cine MRI datasets, obtaining average dice coefficient of 0.84$\pm$0.03 and 0.91$\pm$0.04, respectively. Furthermore, we also demonstrate improved generalization performance of the distance map regularized network on cross-dataset segmentation, showing as much as 42% improvement in myocardium Dice coefficient from 0.56$\pm$0.28 to 0.80$\pm$0.14., 11 pages manuscript, 5 pages supplementary materials

For the liver to remain viable, the resection during hepatectomy procedure should proceed along the major vessels; hence, the resection planes of the anatomic segments are defined, which mark the peripheries of the self-contained segments inside the liver. Liver anatomic segments identification represents an essential step in the preoperative planning for liver surgical resection treatment. The method based on constructing atlases for the portal and the hepatic veins bifurcations, the atlas is used to localize the corresponding vein in each segmented vasculature using atlas matching. Point-based registration is used to deform the mesh of atlas to the vein branch. Three-dimensional distance map of the hepatic veins is constructed; the fast marching scheme is applied to extract the centerlines. The centerlines of the labeled major veins are extracted by defining the starting and the ending points of each labeled vein. Centerline is extracted by finding the shortest path between the two points. The extracted centerline is used to define the trajectories to plot the required planes between the anatomical segments. The proposed approach is validated on the IRCAD database. Using visual inspection, the method succeeded to extract the major veins centerlines. Based on that, the anatomic segments are defined according to Couinaud segmental anatomy. Automatic liver segmental anatomy identification assists the surgeons for liver analysis in a robust and reproducible way. The anatomic segments with other liver structures construct a 3D visualization tool that is used by the surgeons to study clearly the liver anatomy and the extension of the cancer inside the liver.

The objective of this study was to determine the effects of flow rates and layer thicknesses for the aggregates on the estimation of the aggregate gradation of an asphalt mixture based on machine vision. The watershed algorithm based on distance transform and concave point detection algorithm were synthetically used to split the touching particles. Increasing the flow rate overall decreased the prediction accuracy of aggregate gradation by image segmentation; the prediction accuracy decreased with the increase of the layer thickness. The prediction accuracy of the segmentation was better than that of the online detection of the asphalt mixture plant.

In this article, the almost inner (resp. lower, upper) regularity of metrics on a group is proposed. In terms of these regularities, a sufficient and necessary condition for distance transforms to be Lipschitz-1 continuous is given, and relations between balls with different centres and radii are discussed. It turns out that the three almost regularities of metrics introduced and the results obtained here might provide useful tools in discrete analysis, mathematical morphology and image analysis, etc.

A novel variational problem for approximating the distance function (to a domain boundary) is proposed. It is shown that this problem can be efficiently solved by ADMM. A review of several other variational and PDE-based methods for distance function estimation is presented. Advantages of the proposed distance function estimation method are demonstrated by numerical experiments. Applications of the method to the problems of surface curvature estimation and computing the skeleton of a binary image are shown.

A recent trend in optimal motion planning has broadened the research area toward the hybridization of sampling, optimization, and grid-based approaches. A synergy from such integrations can be expected to bring the overall performance improvement, but seamless integration and generalization is still an open problem. In this paper, we suggest a hybrid motion planning algorithm utilizing both sampling and optimization techniques, while simultaneously approximating a configuration-free space. Unlike conventional optimization-based approaches, the proposed algorithm does not depend on a priori information or resolution-complete factors, e.g., a distance field. Ours instead learns spatial information on the fly by exploiting empirical collisions found during the execution, and decentralizes the information over the constructed graph for an efficient reference. With the help of the learned information, we associate the constructed search graph with the approximate configuration-free space so that our optimization-based local planner exploits the local area to identify the connectivity of free space without depending on the precomputed workspace information. To show the novelty of the proposed algorithm, we apply the proposed idea to asymptotic optimal planners with lazy collision checking; lazy PRM$$^*$$ and Batch Informed Tree$$^*$$, and evaluate them against other state-of-the-arts in both synthetic and practical benchmarks with varying degrees of freedom. We also discuss the performance analysis, properties of different algorithm frameworks of lazy collision checking and our approximation method.

Digital reconstruction (tracing) of tree-like structures, such as neurons, retinal blood vessels, and bronchi, from volumetric images and 2D images is very important to biomedical research. Many existing reconstruction algorithms rely on a set of good seed points. The 2D or 3D terminations are good candidates for such seed points. In this paper, we propose an automatic method to detect terminations for tree-like structures based on a multiscale ray-shooting model and a termination visual prior. The multiscale ray-shooting model detects 2D terminations by extracting and analyzing the multiscale intensity distribution features around a termination candidate. The range of scale is adaptively determined according to the local neurite diameter estimated by the Rayburst sampling algorithm in combination with the gray-weighted distance transform. The termination visual prior is based on a key observation-when observing a 3D termination from three orthogonal directions without occlusion, we can recognize it in at least two views. Using this prior with the multiscale ray-shooting model, we can detect 3D terminations with high accuracies. Experiments on 3D neuron image stacks, 2D neuron images, 3D bronchus image stacks, and 2D retinal blood vessel images exhibit average precision and recall rates of 87.50% and 90.54%. The experimental results confirm that the proposed method outperforms other the state-of-the-art termination detection methods.

Various 3D imaging techniques are routinely used to examine biological materials, the results of which are usually a stack of grayscale images. In order to quantify structural aspects of the biological materials, however, they must first be extracted from the dataset in a process called segmentation. If the individual structures to be extracted are in contact or very close to each other, distance-based segmentation methods utilizing the Euclidean distance transform are commonly employed. Major disadvantages of the Euclidean distance transform, however, are its susceptibility to noise (very common in biological data), which often leads to incorrect segmentations (i.e., poor separation of objects of interest), and its limitation of being only effective for roundish objects. In the present work, we propose an alternative distance transform method, the random-walk distance transform, and demonstrate its effectiveness in high-throughput segmentation of three microCT datasets of biological tilings (i.e., structures composed of a large number of similar repeating units). In contrast to the Euclidean distance transform, the random-walk approach represents the global, rather than the local, geometric character of the objects to be segmented and, thus, is less susceptible to noise. In addition, it is directly applicable to structures with anisotropic shape characteristics. Using three case studies—tessellated cartilage from a stingray, the dermal endoskeleton of a starfish, and the prismatic layer of a bivalve mollusc shell—we provide a typical workflow for the segmentation of tiled structures, describe core image processing concepts that are underused in biological research, and show that for each study system, large amounts of biologically-relevant data can be rapidly segmented, visualized, and analyzed.

Background: The number of glomeruli on a kidney biopsy slide followed by glomerular assessment constitute as standard components of a renal pathology report. The prevailing method for glomerular identification and assessment remains manual, labor intensive and non-standardized. In the era of digitized kidney biopsies, an automated method to identify, segment and count glomeruli is highly desirable. Methods and results: We developed an automated method to detect and segment the glomeruli within digitized kidney biopsy images by leveraging a deep learning architecture based on convolutional neural networks (CNN). A total of 275 trichrome-stained images (Average image size: 2560x1920x3 pixels, 1-2 unique images per patient, Scale: 0.85 μm/pixel) processed at 40x magnification from renal biopsies of 171 chronic kidney disease patients treated at the Boston Medical Center from 2009-2012 were analyzed. A sliding window operation was defined to crop each 40x image to smaller images of size 300x300x3 pixels. Each cropped image was then evaluated by clinical experts to identify the presence of a unique glomerulus, and each identified glomerulus was included in the training dataset (n = 751). About the same number of cropped images, containing the non-glomerular regions of the kidney biopsy, served as control cases. The CNN model was constructed as a binary classification problem to discriminate glomerular images from the non-glomerular ones (Performance on test data - Accuracy: 97.47±0.31%; Sensitivity: 96.43±1.89%; Specificity: 98.76±1.44%). Using the trained CNN model, another sliding window operator was developed to scan the digitized biopsies. A heatmap was generated to highlight regions of intensity that the CNN model classified as glomerular regions. Subsequently, two independent image processing strategies, one using steps such as image binarization and image erosion, and the other using image binarization, distance transform and watershed segmentation, were performed on the heatmaps to generate discriminatory signatures of the identified glomeruli. The final step involved automatically drawing a box around the higher intensity areas leading to output images with segmented glomerular regions (Performance on test data - Accuracy: 99.97±0.0086%; Sensitivity: 54.37±0.23%; Specificity: 99.99±0.009%). Conclusion: While used in the context of nephropathology, this study demonstrates the power of artificial intelligence to assess complex histologic structures and identify structural variations. Adoption of such methods of counting and classifying glomeruli using standard histological staining without disturbing the workflow can expedite the assessment of slides by the pathologists and serve as a first step toward more comprehensive automated analysis.

This paper proposes a general method for ab-initio modelling and representation of heterogeneous objects that are associated with complex material variation over complex geometry. Heterogeneous objects like composites and naturally occurring objects (bones, rocks and meteorites) possess multiple and often conflicting properties (like high hardness and toughness simultaneously), which are associated with random and irregular material distribution. Modelling such objects is desired for numerical analysis and additive manufacturing to develop bio-implants, high-performance tools etc. However, it is difficult to define and map the arbitrary material distribution within the object as the material distribution can be independent of the shape parameters or form features used to construct its solid model. This paper represents the source of random and irregular material distribution by mixed-dimensional entities with a focus on modelling compositional heterogeneity. The domain of effect of each material reference entity is defined automatically by using Medial Axis Transform (MAT), where the material distribution can be intuitively prescribed, starting from the material reference entity and terminating at the medial axis segment bounding the corresponding domain. Within such a domain, the spatial variation of the material is captured by a distance field from the material reference entity, which can be controlled locally and independently. These domains are stored using the neighbourhood relation for efficient operations like altering material distribution across the material reference entity and material evaluation for a given geometric location. Results from an implementation for 2.5D objects are shown and the extension to 3D objects is discussed. Highlights Hybrid representation for complex heterogeneous objects. General method for ab-initio modelling of complex heterogeneous objects capture arbitrary material distribution by automatic domain decomposition. Local control of material composition through MAT around material feature. Specification of material composition without adding new entities to shape model.

We propose “convolutional distance transform”—efficient implementations of distance transform. Specifically, we leverage approximate minimum functions to rewrite the distance transform in terms of convolution operators. Thanks to the fast Fourier transform, the proposed convolutional distance transforms have $\mathcal {O}(N\log N)$ complexity, where $N$ is the total number of pixels. The proposed acceleration technique is “distance metric agnostic.” In the special case that the distance function is a $p$ -norm, the distance transform can be further reduced to separable convolution filters; and for Euclidean norm, we achieve $\mathcal {O}(N)$ using constant-time Gaussian filtering.

We develop a fully parallel numerical method which quickly performs 2D and 3D segmentation on GPU to extract anatomical structures from medical images. The algorithm solves the level set equations completely within a Lattice Boltzmann model (LBM). Compared with existing LBM-based segmentation approaches, a parallel distance field regularization is added to the LBM computing scheme to keep computation stable with large time step iteration. This approach also avoids external regularization which has been a major impediment to direct parallelization of level set evolution with LBM. It allows the whole computing process to be efficiently executed on GPU. Moreover, the method can be incorporated with different image features to adopt in various image segmentation tasks. Therefore, our method enables fully GPU accelerated geometric extraction from medical images, leading to high computing performance which is demanded in many practical applications. This method is used to exactly accurate 2D and 3D anatomical structures from many real world CT and MRI images. The achieved results can also directly feed required boundary information to LBM-based hemodynamics simulation.

This paper develops a reliable workflow for multi-objective history matching coupled with a distance-based model selection and ensemble-based data assimilation at a clastic channel reservoir with uncertain geological scenarios. The distance map allocated to each producer determines the probabilities of training images which assess the uncertainty of geological scenarios. K-medoids clustering selects the reservoir models within the ensemble set applied with some training images with less error. These geo-models play as initial ensembles suitable to explain the geological scenarios and ensemble Kalman filter recursively assimilates the oil rates of each producer. The developed workflow, updating reliable reservoir models suitable for well-performance-based history matching, more accurately forecasts water breakthrough and improves the predictability of unknown oil rates with a lower error than those of the conventional ensemble Kalman filter. This framework is able to preserve the spatial characteristics of facies models and reservoir properties without interpreting one fixed scenario. The proposed method can contribute to a reasonable design for data analytics with uncertain geological scenarios and for matching different-scaled well production histories.

Signed distance fields represent objects as distances to the closest surface points with a sign differentiating inside and outside. We present an algorithm to compute a signed distance field from triangle meshes. All data are kept on the GPU, making it ideal for any pure graphics-based context. We split the algorithm into a fast parallel distance transform and a new method of computing the sign. To determine the sign, we compute the winding number for any point using a ray map, a ray-based data structure that preserves geometric meaning while reducing the amount of work to be done for ray tests. Based on that structure, we devise a simple parallel algorithm to sample an exponentially growing number of rays to cope with meshes having deficiencies such as holes or self-intersections. We demonstrate how our method is both fast and able to handle imperfect meshes.

We propose a quantitative analysis method using cell segmentation to evaluate the coating quality of anodic oxide. In order to identify each cell in a titania surface, a boundary region is adaptively detected according to the regional intensity distribution, and a boundary distance map is projected onto the scanning electron microscope (SEM) image of the titania surface. Each cell in the projected image is divided via clustering centered on the local maximum point. The uniformities of the cell distribution, size, and thickness are measured using the angle difference, size difference, and brightness difference between adjacent cells to comprehensively evaluate the quantitative alignment of the titania surface. The proposed method demonstrates the best performance in both cell detection and cell segmentation for SEM images of the titania surface compared with the state-of-the-art methods. The quantitative analysis of the anodic oxide film alignment was verified using color coding, histogram, and quantification of uniformity.

To explore the problem of distance transformations while obstacle existing, this paper presents an obstacle-avoiding Euclidean distance transform method based on cellular automata. This research took the South China Sea as an example, imported the data of land-sea distribution and target points, took the length of the shortest obstacle-avoiding path from current cell to the target cells as the state of a cellular, designed the state transform rule of each cellular that considering a distance operator, then simulated the propagation of obstacle-avoiding distance, and got the result raster of obstacle-avoiding distance transform. After analyzing the effect and precision of obstacle avoiding, we reached the following conclusions:first, the presented method can visually and dynamically show the process of obstacle-avoiding distance transform, can automate calculate the shortest distance bypass the land; second, the method has auto update mechanism, each cellular can rectify distance value according to its neighbor cellular during the simulation process; At last, it provides an approximate solution for exact obstacle-avoiding Euclidean distance transform, the proportional error is less than 3.96%. The proposed method can apply to the fields of shipping routes design, maritime search and rescue, and so on.

In this study, an obstacle avoidance algorithm, with variable range searching, is proposed for a mobile robot. In the case of the distance transform method, the mobile robot cannot avoid the obstacle when it is narrow. We prepare a grid map with the obstacle and distance levels from a destination. A path with both the least number of turns, and the shortest distance, is generated. The search range depends on the distance between the current position and the obstacle. The algorithm can generate the avoidance path without regard to the shape of the obstacle. In the computer simulation, the proposed algorithm can generate a path that avoids the obstacle. As an experiment, a position detection system used contrast differences, by means of a web camera, to get the coordinates of the mobile robot. It is confirmed that, in the experiment, the mobile robot reached the goal grids without hitting the obstacle.

This paper simulates two important processes of batik: crack generation and cloth dyeing. We use FIT algorithm and other methods to reduce the execution time and achieve good visual effects. We attempt to simulate the multi-colored dyeing process without considering the cloth structure. Dyeing of a batik plain weave cloth is simulated by a batik cloth model and a differential diffusion equation. Mottling is simulated by an ellipse model with Perlin noise. Experiments show that these methods can properly simulate the characteristics of batik.

Localization of text from camera captured images with complex background is now-a-days a growing demand of modern IT enable service. Most of the current text localization techniques are sensitive to text features like color, size, style and also to the background clutter. Among all the methods proposed in different literatures, Stroke Filter is much more effective in localization of text. The effectiveness of traditional stroke filter is limited because of its fixed width and is capable of segmenting strokes/texts of predefined range of width. The proposed method uses Fuzzy Distance Transform based adaptive stroke filter which can effectively localize text regions from camera captured images with complex background. The method is applied by experiment on a database containing 600 images and the visual response of text segmentation is quite impressive. To get the accuracy of the proposed method, it is applied on a set of 16 test images and the segmentation result is compared with the ground truth images resulting in a recall, precision and f-measure values of 96.65%, 87.77% and 91.89% respectively.

The morphology and surroundings of cells have been routinely used by pathologists to diagnose the pathological subtypes of gliomas and to assess the malignancy of tumors. Thanks to the advent and development of digital pathology technology, it is possible to automatically analyze whole slides of tissue and focus on the nucleus in order to derive a quantitative assessment that can be used for grading, classification, and diagnosis. During the process of computer-assisted diagnosis, the accurate location and segmentation of nuclei from hematoxylin and eosin (H&E)-stained histopathological images is an important step. In this paper, we proposed a U-Net-based multi-task learning network in which the boundary and region information is utilized to improve the segmentation accuracy of glioma nuclei, especially overlapping ones. To refine the segmentation, a classification model is used to predict the boundary, a regression model is used to predict the distance map, and the final segmentation is obtained by using the fusion layers. The proposed approach was compared with other specially designed boundary-aware methods by using a pathological section dataset that consists of 320 glioma cases from the Huashan Hospital at Fudan University. Both the pixel-level and object-level evaluations showed that the structural modification is effective in segmentation with an F1-score of 0.82, a Hausdorff distance (HD) of 3.95, and an aggregated Jaccard index (AJI) of 0.66 (+0.46%, -3.75%, and +4.09% compared with the unimproved methods, respectively). In addition, comparative experiments on multi-organ nuclei segmentation (MoNuSeg) open dataset proved the advanced nature of the proposed method in the field of nuclei segmentation, especially separating touching objects. The proposed method obtains an AJI of 0.59 and an F1-score of 0.79.

Railway location design is a complex and time-consuming task in which the intricate interactions among railway alignment and station locations must be considered. However, most optimization methods emphasize either alignment optimization or station locations optimization while ignoring the coupling constraints for railway alignment and station locations. In addition, these methods are mainly applied to plain areas with relatively simple topographies and often fail to find feasible alternatives in complex mountainous areas. This paper proposes a concurrent optimization method for railway alignment and station locations. Coupling constraints for railway alignment and station locations are formulated and addressed, and station locations are generated while searching the railway alignment. The method can be applied for more general terrains, including complex mountainous areas. The effectiveness of the method is verified through a real-world case study in a complex mountainous area. The results show that the proposed method can concurrently find the optimal railway alignment and station locations for specific objectives and multiple constraints and provide various promising alternatives.

In low permeability reservoirs, such as shale oil and tight oil, the natural fractures are critical and primary conduit for the flow of oil and gas. The fluid flow is primarily in the fractures and the matrix usually serves as storage. To exploit these unconventional reservoirs efficiently, an optimized well system needs to be built to maximize the profits. Multilateral well is a kind of well with several laterals from one mainbore, which can potentially optimize the production. In this study, we develop a novel multilateral well optimization method for determining the drilling trajectories and configuration in complex fractured reservoirs. To penetrate more fractures for the production improvement, the objective function is to maximize the ratio between the total length of effectively connected fractures to the well and the optimized well length. A connectivity map is constructed based on the topological relationship among discrete fractures such that the region near a large interconnected cluster is given a high value of importance. Then, the fast marching method is employed to generate a distance map based on the connectivity field. The connectivity map will be altered by shading the cluster that has been connected to the well, and the distance map will also be updated so that the next zone of importance can be recognized. Different optimized lateral trajectories and complete configuration are obtained through iterations by updating the connectivity map, while considering lateral interferences. Specific procedures are further demonstrated through an optimization example. Different multilateral well configurations are optimized and compared. Finally, an optimized well configuration is obtained for the specific developed reservoir. The proposed method is verified through a series of benchmark tests of simple geometries and comparison with conventional straight wells. Results show that a quadlateral well can be the most approximate configuration for the fracture system studied. The results of this paper can have implications for well drilling in subsurface energy systems.

Wide-field microscopes are commonly used in neurobiology for experimental studies of brain samples. Available visualization tools are limited to electron, two-photon, and confocal microscopy datasets, and current volume rendering techniques do not yield effective results when used with wide-field data. We present a workflow for the visualization of neuronal structures in wide-field microscopy images of brain samples. We introduce a novel gradient-based distance transform that overcomes the out-of-focus blur caused by the inherent design of wide-field microscopes. This is followed by the extraction of the 3D structure of neurites using a multi-scale curvilinear filter and cell-bodies using a Hessian-based enhancement filter. The response from these filters is then applied as an opacity map to the raw data. Based on the visualization challenges faced by domain experts, our workflow provides multiple rendering modes to enable qualitative analysis of neuronal structures, which includes separation of cell-bodies from neurites and an intensity-based classification of the structures. Additionally, we evaluate our visualization results against both a standard image processing deconvolution technique and a confocal microscopy image of the same specimen. We show that our method is significantly faster and requires less computational resources, while producing high quality visualizations. We deploy our workflow in an immersive gigapixel facility as a paradigm for the processing and visualization of large, high-resolution, wide-field microscopy brain datasets.

In this paper, a multiscale contour steered region integral (MCSRI) method is proposed to classify highly similar shapes with flexible interior connection architectures. A component distance map (CDM) is developed to robustly characterize the flexible interior connection structure, shape of the exterior contour, and their inter-relationship in a shape image. A novel multiscale region transform (MReT) is proposed to perform region integral over different contour-steered strips at all possible scales to effectively integrate patch features, and thus enables a better description of the shape image in a coarse-to-fine manner. It is applied to solve a challenging problem of classifying cultivars from leaf images, which is a new attempt in both biology and computer vision research communities. A soybean cultivar leaf vein database (SoyCultivarVein), which is the first cultivar leaf vein database, is created and presented for performance evaluation. The experimental results demonstrate the superiority of the proposed method over the state-of-the-art methods in similar shape classification and the possibility of cultivar recognition via leaf pattern analysis, which may lead to a new research interest towards fine-level shape analysis on cultivar classification.

We propose an end-to-end method for simultaneously detecting local junctions and global wireframe in man-made environment. Our pipeline consists of an anchor-free junction detection module, a distance map learning module, and a line segment proposing and verification module. A set of line segments are proposed from the predicted junctions with guidance of the learned distance map, and further verified by the proposal verification module. Experimental results show that our method outperforms the previous state-of-the-art wireframe parser by a descent margin. In terms of line segments detection, our method shows competitive performance on standard benchmarks. The proposed networks are end-to-end trainable and efficient.aaThe code will be released on github for reproduction of the results.

The Canny Edge-Based Distance Transform and Hough Transform algorithms were successfully implemented to analyze the size distribution and the orientation of the electrospun PLA/PBS hybrid fiber. The effect of polymer concentration, voltage, feed rate and needle-collector distance were studied. It was found that feed rate of 0.5 ml/h and needle-collector distance of 12 cm is required to generate smooth and uniform hybrid fibers in the smallest size electrospinning with 6 wt.% polymer at voltage of 20 kV. The stationary flat collector with a swaying needle and fast rotating drum disc with a stationary needle were used for the fiber alignment. In order to improve the size distribution and the orientation of the hybrid fibers it is proposed that the fibers should be collected on the rotary drum disc at 700 rpm or higher speed.

The manual methods are often used for measurement of fiber diameter of electrospun membranes, however these methods are not suitable for online quality control. Although number of image analysis methods have been reported for automated measurement of fiber diameter, but there is no literature available regarding reliability, accuracy and simplicity of these measurements. In this study, the distance transform methods suggested by Pourdeyhimi and Ziabari, direct tracking method, DiameterJ software and Zhang method were compared for automated measurement of fiber diameter of electrospun nanofibrous membranes. From one way ANOVA, the difference between the results of fiber diameter was found highly significant. The mean and standard deviation of fiber diameter obtained in case of Ziabari method were extremely close to the results of manual method and also to the true values of the simulated image. Therefore, the more accurate estimation of fiber diameter was confirmed in case of Ziabari method.

When predicting thunderstorms and localized severe weather events, close calls occur more frequently than direct hits. This makes it difficult for traditional verification approaches to fully represent forecast performance since events observed near a forecast event are counted both as misses and false alarms. Timing and location relative to the affected population are therefore two of the most important aspects of such a forecast. Verification of these aspects allows the determination of a safe distance for the user of a given severe weather alert. In this study, we propose a forecast verification measure based on the Generalized Distance Transform that is mathematically rigorous yet intuitive and user-friendly in its interpretation. The proposed measure compares the distance from a user location to an alert area against the distance from the same user location to a set of observed events. Time series for such comparisons can then be constructed, allowing evaluation of the timing error obtained from the difference between the two time series. Finally, the ‘worst overforecast’ and ‘worst underforecast’ are diagnosed in terms of relative distance.

Robot and human interaction has received a significant amount of attention in the robot vision research community in the past decades. This has been motivated by the desire of understanding human gesture/motion tracking and recognition. If you solve tracking problems under the circumstance of fast movement, occlusion, and illumination, then you need to complicate calculation, and hence the computational complex prevents to work in real time. For example, particle filter is an useful algorithm to track objects, even under occlusion and non-rigid motion difference. However, particle filter needs to enough samples to support reliability of the potential candidates of the target. There have done many works in hand tracking. To track hands in real time, Shan(Shan, 2004) made particle filter faster by reducing sample size according to mean shift. On the other hand, Kolsch ( Kolsch&Turk, 2004) designed a fast tracking algorithm that combined Kanade-Lucas-Tomasi(KLT) flocks and k-nearest neighborhood. Some papers concentrated on the particular properties of hands and their features. Non-rigidity of the hand causes difficulties to track because of non-linear dynamics of the articulation. Fei and Reid(Fei&Reid, 2003) dealt with deformation of the hand by constructing two models according to non-rigid motion from rigid motion. HLAC (Higher-Order Local Auto-Correlation) features of Ishihara (Ishihara&Otsu, 2004) achieved efficient information over time domain by Auto-Regressive model. The size of interesting objects is another critical factor for tracking because if its size is too small or changes too fast, object tracking becomes very challenging problem. Franccis(Franccis,2004) dealt with blobs varying their resolution, hence made it possible to track the object with various size in the image sequence. Both hands tracking is simultaneously different from one hand tracking since features such as shape, color etc. between both hands is almost the same each other. Shamaie (Shamaie&Sutherland, 2003) built the model of the movements of bimanual limbs. However, the model needs large enough time to compute distance transform in the image. McAllister( McKenna et al., 2002) solved the both hands tracking by employing contour distance transform and 2D geometric model. In this paper, we propose a new 2D both hands tracking algorithm based on the articulated structure of human body in real time. This method is efficient enough to perform in real time due to the limb model tracking. The model enables to deal with the deformation of hands and nonrigid motion because of the articulate structure of the arm for both hands.