Robust Segmentation Algorithm Research Articles

Robust segmentation and localization of structural planes from photogrammetric point clouds in construction sites

Plane segmentation is a simple yet essential processing step for using 3D point clouds in applications such as temporal data registration and object modeling. The performance of traditional plane segmentation algorithms, especially when addressing photogrammetric point clouds, is significantly limited by factors including the errors of the point positions, noise contamination, and unevenly distributed density, resulting in a low segmentation efficiency and poor segmentation outcome. This paper proposes a robust plane segmentation algorithm for addressing photogrammetric point clouds of construction site scenarios. The implementation of the proposed method includes three major steps: preprocessing, segmentation of the point clouds, and extraction of the planes. The preprocessing is designed to suppress noise and filter outliers, and downsample the dataset. In the subsequent segmentation step, we propose a bottom-up approach utilizing the voxelized structure and facet-based global clustering to obtain planar segments using an automatic and unsupervised clustering of the planar facets selected from the voxelized points. In the last step, the points of the planar segments are extracted using a parametric model-based plane fitting. A 3D Hough transform-based method, constrained by an orientated bounding box, is developed to generate the initial model of the planes. The obtained plane models are then localized by the oriented bounding box and optimized by a local clustering-based plane refinement. The experimental results using synthetic datasets and a low-cluttered laser scanning dataset confirm that the proposed method is superior to the traditional algorithms in key criteria including calculation time, extraction precision, and recall. For real applications on two datasets of different construction sites, the proposed method could segment the majority of the structural planes from the points, with accurate model parameters, sizes, and locations estimated.

Correction propagation for user-assisted optical coherence tomography segmentation: general framework and application to Bruch's membrane segmentation.

Optical coherence tomography (OCT) is a commonly used ophthalmic imaging modality. While OCT has traditionally been viewed cross-sectionally (i.e., as a sequence of B-scans), higher A-scan rates have increased interest in en face OCT visualization and analysis. The recent clinical introduction of OCT angiography (OCTA) has further spurred this interest, with chorioretinal OCTA being predominantly displayed via en face projections. Although en face visualization and quantitation are natural for many retinal features (e.g., drusen and vasculature), it requires segmentation. Because manual segmentation of volumetric OCT data is prohibitively laborious in many settings, there has been significant research and commercial interest in developing automatic segmentation algorithms. While these algorithms have achieved impressive results, the variability of image qualities and the variety of ocular pathologies cause even the most robust automatic segmentation algorithms to err. In this study, we develop a user-assisted segmentation approach, complementary to fully-automatic methods, wherein correction propagation is used to reduce the burden of manually correcting automatic segmentations. The approach is evaluated for Bruch's membrane segmentation in eyes with advanced age-related macular degeneration.

A Local Neighborhood Robust Fuzzy Clustering Image Segmentation Algorithm Based on an Adaptive Feature Selection Gaussian Mixture Model.

Since the fuzzy local information C-means (FLICM) segmentation algorithm cannot take into account the impact of different features on clustering segmentation results, a local fuzzy clustering segmentation algorithm based on a feature selection Gaussian mixture model was proposed. First, the constraints of the membership degree on the spatial distance were added to the local information function. Second, the feature saliency was introduced into the objective function. By using the Lagrange multiplier method, the optimal expression of the objective function was solved. Neighborhood weighting information was added to the iteration expression of the classification membership degree to obtain a local feature selection based on feature selection. Each of the improved FLICM algorithm, the fuzzy C-means with spatial constraints (FCM_S) algorithm, and the original FLICM algorithm were then used to cluster and segment the interference images of Gaussian noise, salt-and-pepper noise, multiplicative noise, and mixed noise. The performances of the peak signal-to-noise ratio and error rate of the segmentation results were compared with each other. At the same time, the iteration time and number of iterations used to converge the objective function of the algorithm were compared. In summary, the improved algorithm significantly improved the ability of image noise suppression under strong noise interference, improved the efficiency of operation, facilitated remote sensing image capture under strong noise interference, and promoted the development of a robust anti-noise fuzzy clustering algorithm.

Robust Self-Sparse Fuzzy Clustering for Image Segmentation

Traditional fuzzy clustering algorithms suffer from two problems in image segmentations. One is that these algorithms are sensitive to outliers due to the non-sparsity of fuzzy memberships. The other is that these algorithms often cause image over-segmentation due to the loss of image local spatial information. To address these issues, we propose a robust self-sparse fuzzy clustering algorithm (RSSFCA) for image segmentation. The proposed RSSFCA makes two contributions. The first concerns a regularization under Gaussian metric that is integrated into the objective function of fuzzy clustering algorithms to obtain fuzzy membership with sparsity, which reduces a proportion of noisy features and improves clustering results. The second concerns a connected-component filtering based on area density balance strategy (CCF-ADB) that is proposed to address the problem of image over-segmentation. Compared to the integration of local spatial information into the objective functions, the presented CCF-ADB is simpler and faster for the removal of small areas. Experimental results show that the proposed RSSFCA addresses two problems in current fuzzy clustering algorithms, i.e., the outlier sensitivity and the over-segmentation, and it provides better image segmentation results than state-of-the-art algorithms.

Research on a Single-Tree Point Cloud Segmentation Method Based on UAV Tilt Photography and Deep Learning Algorithm

Developing a robust point cloud segmentation algorithm for individual trees from an amount of point cloud data has great significance for tracking tree changes. This method can measure the size, growth, and mortality of individual trees to track and understand forest carbon storage and variation. Traditional measurement methods are not only slow but also tardy. In order to obtain forest information better and faster, this article focuses on two aspects: The first is using UAVs to obtain multiview remote sensing images of the forest, and then using the structure from motion algorithm to construct the forest sparse point cloud and patch-based MVS algorithm to construct the dense point cloud. The second is that a targeted point cloud deep learning method is proposed to extract the point cloud of a single tree. The research results show that the accuracy of single-tree point cloud segmentation of deep learning methods is more than 90%, and the accuracy is far better than traditional planar image segmentation and point cloud segmentation. The combination of point cloud data acquisition with UAV remote sensing and point cloud deep learning algorithms can meet the needs of forestry surveys. It is undeniable that this method, as a forestry survey tool, has a large space for promotion and possible future development.

Robust image segmentation for feature extraction from internal combustion engine in-cylinder images

In-cylinder imaging diagnostics for internal combustion engines provide rich information on the structure and evolution of reaction zone features, which affect both engine out emissions and efficiency. However, the most common analysis of in-cylinder combustion luminosity imaging considers ensemble averaged images, which are not suitable for characterizing processes that vary significantly between cycles, such as ignition and soot formation and oxidation. Here, a robust image segmentation algorithm is presented for feature extraction from single-cycle in-cylinder combustion images and is used with a ‘combination of interpretations’ (COI) approach to analyze OH*-chemiluminescence imaging of premixed and non-premixed natural gas combustion modes in an optically-accessible reciprocating engine.Dynamic thresholding and region size filtering are combined with watershed segmentation to create a parameterized adaptive watershed (PAW) segmentation algorithm. The fusion of these segmentation methods is novel to combustion imaging and is demonstrated to provide quantified improvement relative to the current state of the art segmentation methods; PAW segmentation provides increased sensitivity for early ignition processes, and more robustly identifies the reaction zones at later stages of combustion. The PAW algorithm requires no adjustment between the two considered combustion modes or for any stage of the combustion process. The reliability of the PAW output enables feature extraction of individual reaction zone location and area from the combustion images using a polar-sector coordinate system for COI analysis. This approach characterizes the cyclic variability of individual fuel jets, identifies coupling of auto-ignition behavior between adjacent reaction zones, and demonstrates systematic errors arising from measurement of auto-ignition in ensemble averaged images. Application of PAW segmentation and the analysis approach presented here can provide more complete characterization of other spatially-resolved internal combustion diagnostics, particularly where there is high process variability, overlapping image regions, or wide signal intensity ranges.

Automatic volumetry of cerebrospinal fluid and brain volume in severe paediatric hydrocephalus, implementation and clinical course after intervention.

In childhood hydrocephalus, both the amount of cerebrospinal fluid and the brain volume are relevant for the prognosis of the development and for therapy monitoring. Since classical planar measurements of ventricular size are subject to strong limitations, imprecise and neglect brain volume, 3D volumetry is most desirable. We used and evaluated the robust segmentation algorithms of the freely available FSL-toolbox in paediatric hydrocephalus patients before and after specific therapy. Retrospectively 76 pre- and postoperative high-resolution T2-weighted MRI sequences (true FISP, 1 mm isovoxel) were analyzed in 38 patients with paediatric hydrocephalus (mean 4.4 ± 5.1 years) who underwent surgical treatment (ventriculo-peritoneal (VP) shunt n = 22, endoscopic third ventriculostomy (ETV) n = 16). After preprocessing, the 3D-datasets were skull stripped to estimate the inner skull surface. Following, a 2 class segmentation into different tissue types (brain matter and CSF) was performed. The volumes of CSF and brain were calculated. The method could be implemented in an automated fashion in all 76 MRIs. In the VP shunt cohort, the amount of CSF (p < 0.001) decreased. Consecutively brain volume increased significantly (p < 0.001). Following ETV, CSF volume (p = 0.019) decreased significantly (p = 0.012) although the reduction was less pronounced than after shunt implantation. Brain volume expanded (p = 0.02). A reliable automated segmentation of CSF and brain could be performed with the implemented algorithm. The method was able to track changes after therapy and detected significant differences in CSF and brain volumes after shunting and after ETV.

Segmentation and Diagnosis of Liver Carcinoma Based on Adaptive Scale-Kernel Fuzzy Clustering Model for CT Images.

Medical image analysis plays an important role in computer-aided liver-carcinoma diagnosis. Aiming at the existing image fuzzy clustering segmentation being not suitable to segment CT image with non-uniform background, a fast robust kernel space fuzzy clustering segmentation algorithm is proposed. Firstly, the sample in euclidean space is mapped into the high dimensional feature space through the kernel function. Then the linear weighted filtering image is obtained by combining the current pixel with its neighborhood pixels through the space information in CT image. Finally, the two-dimensional histogram between the clustered pixel and its neighborhood mean is introduced into the robust kernel space image fuzzy clustering, and the iterative expression of the fast robust fuzzy clustering in kernel space is obtained by using Lagrange multiplier method. The experimental results on four databases show that our proposed method can segment liver tumors from abdominal CT volumes effectively and automatically, and the comprehensive segmentation performance of the proposed method is superior to that of several existing methods.

An Accurate and Robust Region-Growing Algorithm for Plane Segmentation of TLS Point Clouds Using a Multiscale Tensor Voting Method

The accuracy and robustness of plane segmentation using a region-growing algorithm remains an important and challenging topic for terrestrial laser scanning point clouds. The plane segmentation of a region-growing algorithm depends heavily on the seed point, as there are currently no universally valid criteria. This article proposes a multiscale tensor voting method (MSTVM) to determine the appropriate seed point for the region-growing algorithm. A comprehensive plane strength indicator calculated by the semivariogram model has been established to assess whether a certain point is suitably considered as a seed point or not. A point cloud containing 17, 881 points in a 400-m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> area was selected to validate the proposed algorithm. The results suggest that the scale range calculated by the semivariogram model can effectively mitigate the scale effect of the tensor voting method (TVM). The comprehensive plane strength of our proposed algorithm in seed point determination is shown to be more salient than the principal component analysis and the TVM. The findings further reveal that the utility of the MSTVM-based region-growing algorithm can achieve more accurate plane segmentation results and perform with better robustness in noisy point clouds. This allows our proposed method to be more widely applied to complex real situations.

An image segmentation algorithm based on double-layer pulse-coupled neural network model for kiwifruit detection

Finding a universal and accurate image segmentation algorithm for kiwifruit detection under varying illumination and complex background has become one of the most challenging problems in machine vision research. In this study, a robust segmentation algorithm based on a double-layer pulse-coupled neural network (PCNN) model is proposed. First of all, an improved PCNN merged with the image frequency-tuned saliency is devised as a basic structure. Secondly, in the red-green-blue color mode, the optimal color-difference information of a kiwifruit image is determined in the first layer of this double-layer PCNN. Then, enhanced hue features are fused with these optimal color-difference features by the total variation model. Finally, the target regions are built by the re-segmentation of the second layer of this double-layer PCNN. Experimental results demonstrate that the proposed algorithm significantly outperforms the typical existing algorithms in terms of the subjective visual effect and the objective quantitative evaluation.

4D Light Field Superpixel and Segmentation.

Superpixel segmentation of 2D images has been widely used in many computer vision tasks. Previous algorithms model the color, position, or higher spectral information for segmenting a 2D image. However, limited to the Gaussian imaging principle in a traditional camera, where each pixel is formed by summing lots of light rays from different angles, there is not a thorough segmentation solution to eliminate the ambiguity in defocus and occlusion boundary areas. In this paper, we consider the essential element of image pixel, i.e., rays in light space, and propose light field superpixel (LFSP) to eliminate the ambiguity. The LFSP is first defined mathematically and then two evaluation metrics, named LFSP self-similarity and effective label ratio, are proposed to evaluate the refocus-invariant and full-sliced properties of segmentation. By building a clique system containing 80 neighbors in light field, a robust refocus-invariant LFSP segmentation algorithm is developed. Experimental results on both synthetic and real light field datasets demonstrate the advantages over the current state of the art in terms of traditional evaluation metrics. Additionally, the LFSP self-similarity evaluations under different light field refocus levels show the refocus-invariance of the proposed algorithm. The full-sliced property of the proposed LFSP algorithm is verified by comparing it with the classical supervoxel algorithms. Finally, an LFSP-based application is demonstrated to show the effectiveness of LFSP in light field editing.

Segmentation of bias field induced brain MR images using rough sets and stomped-t distribution

Over the past few decades, automatic segmentation of brain magnetic resonance (MR) images into different tissue classes has remained an important research area, particularly due to the presence of bias field artifact in MR images. In this regard, the stomped normal (StN) distribution is proved to generate an optimal representation of the intensity distribution in brain MR images, by incorporating the properties of rough sets in the probabilistic framework. The StN distribution is capable of successfully modelling the central tendency, dispersion, and width of the intensity distribution. However, it does not take into consideration the kurtosis of the distribution, which controls the concentration of values around the mean and shape of the tail of intensity distribution. In this regard, the paper presents a novel method for simultaneous segmentation and bias field correction in brain MR images. It integrates the concept of rough sets and the merit of a recently introduced probability distribution, called stomped-t (St-t) distribution. The St-t distribution incorporates the property of kurtosis in rough-probabilistic framework, where each tissue class is modelled using a crisp lower approximation and a probabilistic boundary region. The brain MR image is modelled using a mixture of finite number of St-t distributions and one uniform distribution. The uniform distribution takes into account cerebro-spinal fluid, pathologies, and other non-brain tissues. The proposed method employs both expectation-maximization algorithm and hidden Markov random field model for accurate and robust image segmentation. The performance of the proposed approach, along with a comparison with related methods, is illustrated on a set of synthetic and real brain MR images for different bias fields and noise levels.

Automatic Brain Tumor Segmentation Method Based on Modified Convolutional Neural Network.

In the domain of brain diseases, it is difficult for image registration after some brain structures are severely deformed because of diseases. Fortunately, convolutional neural network have gained many promising results in semantic segmentation challenging tasks in recent years. To enhance the performance of automatic brain tumor segmentation, this paper presents a robust segmentation algorithm based on convolutional neural network, which achieved improvement of 3.84% in segmenting the enhancing tumor. Our network architecture is developed from the prevalent U-Net. We combined it with ResNet and modified it to maximize its performance in our brain tumor segmentation task. In this work, BraTS 2017 dataset was employed to train and test the proposed network. Data imbalance was dealt with using a weighted cross entropy loss function. The problem of overfitting was handled through data augmentation. The proposed method achieved averaged dice scores of 0.883, 0.781 and 0.748 for whole tumor, tumor core and enhancing tumor respectively in the validation set and 0.877, 0.774, 0.757 respectively in the testing set.

A fully-automated, robust, and versatile algorithm for long-term budding yeast segmentation and tracking.

Live cell time-lapse microscopy, a widely-used technique to study gene expression and protein dynamics in single cells, relies on segmentation and tracking of individual cells for data generation. The potential of the data that can be extracted from this technique is limited by the inability to accurately segment a large number of cells from such microscopy images and track them over long periods of time. Existing segmentation and tracking algorithms either require additional dyes or markers specific to segmentation or they are highly specific to one imaging condition and cell morphology and/or necessitate manual correction. Here we introduce a fully automated, fast and robust segmentation and tracking algorithm for budding yeast that overcomes these limitations. Full automatization is achieved through a novel automated seeding method, which first generates coarse seeds, then automatically fine-tunes cell boundaries using these seeds and automatically corrects segmentation mistakes. Our algorithm can accurately segment and track individual yeast cells without any specific dye or biomarker. Moreover, we show how existing channels devoted to a biological process of interest can be used to improve the segmentation. The algorithm is versatile in that it accurately segments not only cycling cells with smooth elliptical shapes, but also cells with arbitrary morphologies (e.g. sporulating and pheromone treated cells). In addition, the algorithm is independent of the specific imaging method (bright-field/phase) and objective used (40X/63X/100X). We validate our algorithm’s performance on 9 cases each entailing a different imaging condition, objective magnification and/or cell morphology. Taken together, our algorithm presents a powerful segmentation and tracking tool that can be adapted to numerous budding yeast single-cell studies.

Unsupervised Deep Learning for Bayesian Brain MRI Segmentation.

Probabilistic atlas priors have been commonly used to derive adaptive and robust brain MRI segmentation algorithms. Widely-used neuroimage analysis pipelines rely heavily on these techniques, which are often computationally expensive. In contrast, there has been a recent surge of approaches that leverage deep learning to implement segmentation tools that are computationally efficient at test time. However, most of these strategies rely on learning from manually annotated images. These supervised deep learning methods are therefore sensitive to the intensity profiles in the training dataset. To develop a deep learning-based segmentation model for a new image dataset (e.g., of different contrast), one usually needs to create a new labeled training dataset, which can be prohibitively expensive, or rely on suboptimal ad hoc adaptation or augmentation approaches. In this paper, we propose an alternative strategy that combines a conventional probabilistic atlas-based segmentation with deep learning, enabling one to train a segmentation model for new MRI scans without the need for any manually segmented images. Our experiments include thousands of brain MRI scans and demonstrate that the proposed method achieves good accuracy for a brain MRI segmentation task for different MRI contrasts, requiring only approximately 15 seconds at test time on a GPU.

Fusion noise-removal technique with modified dark-contrast algorithm for robust segmentation of acute leukemia cell images

Segmentation is the major area of interest in the field of image processing stage. In an automatic diagnosis of acute leukemia disease, the crucial process is to achieve the accurate segmentation of acute leukemia blood image. Generally, there are three requirements of image segmentation for medical purposes, namely; accuracy, robustness and effectiveness which have received considerable critical attention. As such, we propose a new (modified) dark contrast enhancement technique to enhance and automatically segment the acute leukemic cells. Subsequently, we used a fusion 7 × 7 median filter as well as the seeded region growing area extraction (SRGAE) algorithm to minimise the salt-and-pepper noise, apart from preserving the post-segmentation edge. As per the outcomes, the accuracy, sensitivity, and specificity of this method were 91.02%, 83.68%, and 91.57% respectively.

Fully automatic detection and segmentation of abdominal aortic thrombus in post-operative CTA images using Deep Convolutional Neural Networks

Computerized Tomography Angiography (CTA) based follow-up of Abdominal Aortic Aneurysms (AAA) treated with Endovascular Aneurysm Repair (EVAR) is essential to evaluate the progress of the patient and detect complications. In this context, accurate quantification of post-operative thrombus volume is required. However, a proper evaluation is hindered by the lack of automatic, robust and reproducible thrombus segmentation algorithms. We propose a new fully automatic approach based on Deep Convolutional Neural Networks (DCNN) for robust and reproducible thrombus region of interest detection and subsequent fine thrombus segmentation. The DetecNet detection network is adapted to perform region of interest extraction from a complete CTA and a new segmentation network architecture, based on Fully Convolutional Networks and a Holistically-Nested Edge Detection Network, is presented. These networks are trained, validated and tested in 13 post-operative CTA volumes of different patients using a 4-fold cross-validation approach to provide more robustness to the results. Our pipeline achieves a Dice score of more than 82% for post-operative thrombus segmentation and provides a mean relative volume difference between ground truth and automatic segmentation that lays within the experienced human observer variance without the need of human intervention in most common cases.

A segmentation of pulmonary nodules based on improved fuzzy C-means clustering algorithm

According to reports, lung cancer is gradually becoming the first cancer that threatens human life. The early stage of lung cancer is in the form of pulmonary nodules. The key issue in computer-aided diagnosis of lung tumors is to correct and accelerate rapid segmentation of diseased tissue. Therefore, this paper proposes a robust fuzzy c-mean clustering algorithm for pulmonary nodules segmentation, which can effectively improve the adaptive degree of local domain pixels. Since the information of the domain pixels does not necessarily have a positive correlation with the central pixels, the reference mechanism of domain window pixel information needs to be redefined. The robust fuzzy c-means clustering algorithm redefines the grayscale of the spatial pixel points in the domain and selects different fuzzy factors according to the reference standard. Based on this, the weights of different fuzzy factors are updated according to the characteristics of pixel points and gray fluctuation in pixel domain. The experimental results show that this method is superior to other typical algorithms in the segmentation of pulmonary nodules.

A Multi-Anatomical Retinal Structure Segmentation System for Automatic Eye Screening Using Morphological Adaptive Fuzzy Thresholding

Eye exam can be as efficacious as physical one in determining health concerns. Retina screening can be the very first clue for detecting a variety of hidden health issues including pre-diabetes and diabetes. Through the process of clinical diagnosis and prognosis; ophthalmologists rely heavily on the binary segmented version of retina fundus image; where the accuracy of segmented vessels, optic disc, and abnormal lesions extremely affects the diagnosis accuracy which in turn affect the subsequent clinical treatment steps. This paper proposes an automated retinal fundus image segmentation system composed of three segmentation subsystems follow same core segmentation algorithm. Despite of broad difference in features and characteristics; retinal vessels, optic disc, and exudate lesions are extracted by each subsystem without the need for texture analysis or synthesis. For sake of compact diagnosis and complete clinical insight, our proposed system can detect these anatomical structures in one session with high accuracy even in pathological retina images. The proposed system uses a robust hybrid segmentation algorithm combines adaptive fuzzy thresholding and mathematical morphology. The proposed system is validated using four benchmark datasets: DRIVE and STARE (vessels), DRISHTI-GS (optic disc), and DIARETDB1 (exudates lesions). Competitive segmentation performance is achieved, outperforming a variety of up-to-date systems and demonstrating the capacity to deal with other heterogeneous anatomical structures.

Panicle-SEG: a robust image segmentation method for rice panicles in the field based on deep learning and superpixel optimization

BackgroundRice panicle phenotyping is important in rice breeding, and rice panicle segmentation is the first and key step for image-based panicle phenotyping. Because of the challenge of illumination differentials, panicle shape deformations, rice accession variations, different reproductive stages and the field’s complex background, rice panicle segmentation in the field is a very large challenge.ResultsIn this paper, we propose a rice panicle segmentation algorithm called Panicle-SEG, which is based on simple linear iterative clustering superpixel regions generation, convolutional neural network classification and entropy rate superpixel optimization. To build the Panicle-SEG-CNN model and test the segmentation effects, 684 training images and 48 testing images were randomly selected, respectively. Six indicators, including Qseg, Sr, SSIM, Precision, Recall and F-measure, are employed to evaluate the segmentation effects, and the average segmentation results for the 48 testing samples are 0.626, 0.730, 0.891, 0.821, 0.730, and 76.73%, respectively. Compared with other segmentation approaches, including HSeg, i2 hysteresis thresholding and jointSeg, the proposed Panicle-SEG algorithm has better performance on segmentation accuracy. Meanwhile, the executing speed is also improved when combined with multithreading and CUDA parallel acceleration. Moreover, Panicle-SEG was demonstrated to be a robust segmentation algorithm, which can be expanded for different rice accessions, different field environments, different camera angles, different reproductive stages, and indoor rice images. The testing dataset and segmentation software are available online.ConclusionsIn conclusion, the results demonstrate that Panicle-SEG is a robust method for panicle segmentation, and it creates a new opportunity for nondestructive yield estimation.