Connected Components Research Articles

BINYAS: a complex document layout analysis system

Document layout analysis (DLA) is an irreplaceable pre-requisite for the development of a comprehensive document image processing and analysis system. The main purpose of DLA is to segment an input document image into its constituent and coherent regions and identify their classes. In this paper, we propose a competent DLA system, named as BINYAS, based on the connected component (CC) and pixel analysis based approach. Here, we initially identify the regions and then classify these regions as paragraph, separator, graphic, image, table, chart, and inverted text etc. The proposed system is evaluated on four publicly available standard datasets, namely ICDAR 2009, 2015, 2017 and 2019 page segmentation competition datasets, and the performance is compared with many contemporary methods, which also include some well-known software products and deep learning based methods. Experimental results show that our method performs significantly better than state-of-the-art methods in terms of the evaluation metrics considered by the research community of this domain.

Efficient segmentation of lumbar intervertebral disc from MR images

Segmentation of spine Magnetic Resonance Images (MRIs) has become an indispensable process in the diagnosis of lumbar disc degeneration, causing low back pain. Over the last decade of years, computer-directed diagnosis of disease, as well as computer-guided spine surgery, is based on the two-dimensional (2D) analysis of mid-sagittal slice of MRI. This work proposes an automatic strategy to extract the 3D segmentation of the normal disc as well as degenerated lumbar intervertebral discs (IVDs) from T2-weighted Turbo Spin Echo MRI of the spine using Connected Component (CC) analysis algorithm and statistical shape analysis. The challenges faced by the IVD segmentation includes (i) partial volume effects (ii) intensity inhomogeneity (iii) grey level overlap of different soft tissues. The proposed method first pre-processes the dataset and enables it for the application of the CC algorithm. The CC (subsets of pixels of the disc) of the spine MRI is extracted and apply statistical shape analysis for the refinement of the segmentation results to detect IVDs. Experimental results of the proposed method show a robust segmentation, accomplishing the dice similarity index of 92.4% and thus achieving a low error rate. Other performance measures such as Precision, Accuracy, JaccardIdx, JaccardDist, Global Consistency Error, Variation of Information, etc were also evaluated. The algorithm is evaluated quantitatively using adequate experiments on a dataset of 15 MRI scans, of different scenarios such as healthy and degenerate disc and this proposed method is verified as a promising accurate method for the automatic segmentation of IVD.

A Multi-Agent Approach to Segment Arabic Handwritten Text Lines

In text line segmentation, there are three classes of methods: either by sorting physical units such as pixels or connected components (CC) constituting a line or by searching for the baseline of each word and grouping together those who participate in the same line. The third class analyzes the separation locations between the lines. After an overview of lines segmentation approaches, the authors introduced a new method emphasizing its simplicity, speed, and originality. The proposed approach detects the starting components of the lines in the first step. In the second step, it defines a number of agents that start the segmentation process from their starting points between the starting components of lines. Each agent aims to reach the left edge of the document through the correct path. The algorithm used by the agents is based on the morphological process, characteristics of the Arabic manuscript and a communication system. The experimental results on an Arabic dataset show that this approach is an effective solution for the segmentation of lines from different Arabic manuscripts.

Handwritten Text Segmentation via End-to-End Learning of Convolutional Neural Networks

We present a method that separates handwritten and machine-printed components that are mixed and overlapped in documents. Many conventional methods addressed this problem by extracting connected components (CCs) and classifying the extracted CCs into two classes. They were based on the assumption that two types of components are not overlapping each other, while we are focusing on more challenging and realistic cases where the components are often overlapping each other. For this, we propose a new method that performs pixel-level classification with a convolutional neural network. Unlike conventional neural network methods, our method works in an end-to-end manner and does not require any preprocessing steps (e.g., foreground extraction, handcrafted feature extraction, and so on). For the training of our network, we develop a cross-entropy based loss function to alleviate the class imbalance problem. Regarding the training dataset, although there are some datasets of mixed printed characters and handwritten scripts, most of them do not have overlapping cases and do not provide pixel-level annotations. Hence, we also propose a data synthesis method that generates realistic pixel-level training samples having many overlappings of printed and handwritten components. Experimental results on synthetic and real images have shown the effectiveness of the proposed method. Although the proposed network has been trained only with synthetic images, it also improves the OCR rate of real documents. Specifically, the OCR rate for machine-printed texts is increased from 0.8087 to 0.9442 by removing the overlapped handwritten scribbles by our method.

Probabilistic homogeneity for document image segmentation

In this paper we propose a novel probabilistic framework for document segmentation exploiting human perceptual recognition of text regions from complicated layouts. In particular, we conceptualize text homogeneity as the Gestalt pattern displayed in text regions, characterized by proximately and symmetrically arranged units with similar morphological and texture features. We model this pattern in the local region of a connected component (CC) using an hierarchical formulation, which simulates a random walk-and-check on a graph encoding the neighborhood of the CC. The proposed formulation allows an effective computation of what we call the probabilistic local text homogeneity (PLTH) using a weighted summation of the weights of the graph, which are derived from a probabilistic description of the homogeneity between neighboring CCs and computed through Bayesian cue integration. The proposed PLTH enables a multi-aspect analysis, where various primitives such as geometrical configuration, morphological features, texture characterization and location priors are integrated in one computational probabilistic model. This enables an effective text and non-text classification of CCs preceding any grouping process, which is currently absent in document segmentation. Experimental results show that our segmentation method based on the proposed PLTH model improves upon the state-of-the-art.

PACC: Large scale connected component computation on Hadoop and Spark.

A connected component in a graph is a set of nodes linked to each other by paths. The problem of finding connected components has been applied to diverse graph analysis tasks such as graph partitioning, graph compression, and pattern recognition. Several distributed algorithms have been proposed to find connected components in enormous graphs. Ironically, the distributed algorithms do not scale enough due to unnecessary data IO & processing, massive intermediate data, numerous rounds of computations, and load balancing issues. In this paper, we propose a fast and scalable distributed algorithm PACC (Partition-Aware Connected Components) for connected component computation based on three key techniques: two-step processing of partitioning & computation, edge filtering, and sketching. PACC considerably shrinks the size of intermediate data, the size of input graph, and the number of rounds without suffering from load balancing issues. PACC performs 2.9 to 10.7 times faster on real-world graphs compared to the state-of-the-art MapReduce and Spark algorithms.

Connected Component Based Word Spotting on Persian Handwritten image documents

Word spotting is to make searchable unindexed image documents by locating word/words in a doc-ument image, given a query word. This problem is challenging, mainly due to the large numberof word classes with very small inter-class and substantial intra-class distances. In this paper, asegmentation-based word spotting method is presented for multi-writer Persian handwritten doc-uments using attribute-based classi cation and label-embedding. For this purpose, a hierarchicalframework is proposed, in which at rst, the candidate are selected based on connected compo-nents(CCs) sequence. Then, the query word is segmented to constructor CCs, and similar CCs countin the candidate region of document are selected based on their distances to the CCs count of thequery word. As a result, the candidate regions are extracted. In the nal phase, the query wordis located only in the candidate regions of the document. A well known Persian handwritten textdataset, namely FTH, is chosen as a benchmark for the presented method. The results shows thatthe proposed method outperforms the state-of-the-art methods, 81.02 percent for unseen word classretrieval.

Connected filters on generalized shape-Spaces

Classical hierarchical image representations and connected filters work on sets of connected components (CC). These approaches can be defective to describe the relations between disjoint objects or partitions of images. In practice, objects can be made of several connected components in images (due to occlusions for example), therefore it can be interesting to be able to take into account the relationship between these components to be able to detect the whole object. In Mathematical Morphology, second-generation connectivity (SGC) and tree-based shape-spaces study this relation between the connected components of an image. However, they have limitations. For this reason, we propose in this paper an extension of the usual shape-space paradigm into what we call a Generalized Shape-Space (GSS). This new paradigm allows us to analyze any graph of connected components hierarchically and to filter them thanks to connected operators.

Real-time localization of multi-oriented text in natural scene images using a linear spatial filter

This paper proposes a multi-oriented text localization method in natural images suitable for real-time processing of high-definition video on portable and mobile devices. Our method is based on the connected components (CC) approach: first, CC are isolated by convolving a multi-scale pyramid with a specifically designed linear spatial filter followed by hysteresis thresholding. Next, non-textual CC are pruned employing a local classifier consisting of a cascade of multilayer perceptron (MLP) fed with increasingly extended feature vectors. The stroke width feature is estimated in linear time complexity by computing the maximal inscribed squares in the CC. Candidate CC and their neighbors are then checked using a more context aware neural network classifier that takes into account the target CC and their vicinity. Finally, text sequences are extracted in all pyramid levels and fused using dynamic programming. The main contribution of the work presented here is execution speed: the CPU-only parallel implementation of the proposed method is capable of processing 1080p HD video at nearly 30 frames per second on a standard laptop. Furthermore, when benchmarked on the ICDAR 2013 Robust Reading and on the ICDAR 2015 Incidental Scene Text data sets, our system performs more than twice faster than the state-of-the-art, while still delivering competitive results in terms of precision and recall.

Matrix logarithms and range of the exponential maps for the symmetry groups , and the Lorentz group

Physicists know that covering the continuously connected component of the Lorentz group can be achieved through two Lie algebra exponentials, whereas one exponential is sufficient for compact symmetry groups like SU(N) or SO(N). On the other hand, both the general Baker-Campbell-Hausdorff formula for the combination of matrix exponentials in a series of higher order commutators, and the possibility to define the logarithm of a general matrix through the Jordan normal form, seem to naively suggest that even for non-compact groups a single exponential should be sufficient. We provide explicit constructions of for all matrices in the fundamental representations of the non-compact groups , and SO(1, 2). The construction for also yields logarithms for SO(1, 3) through the spinor representations. However, it is well known that single Lie algebra exponentials are not sufficient to cover the Lie groups and . Therefore we revisit the maximal neighbourhoods and which can be covered through single exponentials with or , respectively, to clarify why or outside of the corresponding domains or . On the other hand, for the Lorentz groups SO(1, 2) and SO(1, 3), we confirm through construction of the logarithm that every transformation in the connectivity component of the identity element can be represented in the form with or , respectively. We also examine why the proper orthochronous Lorentz group can be covered by single Lie algebra exponentials, whereas this property does not hold for its covering group : The logarithms in correspond to logarithms on the first sheet of the covering map , which is contained in . The special linear groups and the Lorentz group therefore provide instructive examples for different global behaviour of non-compact Lie groups under the exponential map.

Comparative Study and Proof of Single-Pass Connected Components Algorithms

Union-find algorithms form the basis of managing sets of equivalent labels within most connected components labelling algorithms. The new class of single-pass connected components analysis (CCA) algorithms (where a feature vector of each component is extracted during processing) are analysed and compared within this context. Such algorithms have been developed for stream processing, using customised hardware architectures. Many of these use an improved union-find algorithm requiring only a single lookup for its find operation. This paper analyses this optimisation and formally proves that the resulting single lookup connected components algorithm associates each pixel with its correct component when extracting the components’ feature vectors. Analysis of the algorithm led to a new double lookup algorithm that reduces the total number of memory accesses and is a step towards unifying pixel-based methods and run-based methods. State-of-the-art CCA algorithms are compared in terms of the number of memory accesses, which is a limiting factor for hardware-based acceleration, with key implementation trade-offs identified between hardware resources and worst-case processing speed.

MAGNETIC-PULSE PRESSING OF ELECTRICAL CONNECTIONS FOR STRANDED WIRES

Purpose. The research of the process peculiarities of magnetic- pulse fitting of electric stranded conductor joints, made of different materials, using couplings. Evaluation of loading optimal parameters, providing high operational reliability of electric connecting units. Methodology. In order to carry out simulation and research of the process of magnetic-pulse fitting of electric stranded conductor joints CRUG24 software package was used, which was developed to estimate impact interaction. Handling the problem was carried out numerically using finite differences. Metallographic study of collected cross-sectional cuts was performed with the use of optical microscope МЕТАМ ЛВ-71, equipped with digital-still camera, connected to the computer, which used image analysis system IMEGE Expert Pro3. The electrical tests of wire joints were carried out using the thermal bench from exposure to two factors: heating with rated current and expansion by operating load. Results. It was ascertained that magnetic-pulse pressing of electric joints was followed by partial self-purification and bedding component contacting surfaces of electric joints. Oxides and contaminating impurities were expelled into small localized zones between wires, between a wire and a coupling, which resulted in the contact of juvenile surfaces. Upon mutual deformation and displacement of metal wire surface capacity size and coupling tight mechanical contact was created, which provided minimal transient resistivity. The existence of residual compression stress provides the longstanding high-quality electric contact in joints. While using magnetic-pulse pressing of electric joints, due to high speed of deformation and impact of great inertial forces, deformation containment of connected components takes place in the zone of load action. The wires in contact with each other and with couplings generate faceting, but coupling sidewall hardly has any thinning. Filled density of cross-section is approximately 100 %. This fact provides a high degree of sealing capacity of joints, which to a wide extent prevents the oxidation of contacting surfaces and the rise of transient resistivity of electric joints in the operational process. As a result of processing of the results of thermal and electrical tests it was ascertained that pressed joint factors of defectiveness with all types of wires according to thermal impact and resistivity, are significantly lower than unity. Practical value. Magnetic-pulse pressing of unattended joints in electric stranded monometallic, bimetallic and composite conductors provides high operational reliability of connecting units and it may be used while mounting overhead system of railways, transmission lines and fitting connecting components of electrical transport electric circuits.

A Shared-Memory Algorithm for Updating Tree-Based Properties of Large Dynamic Networks

This paper presents a network-based template for analyzing large-scale dynamic data. Specifically, we propose a novel shared-memory parallel algorithm for updating tree-based structures or properties, such as connected components (CC) and minimum spanning trees (MST), on dynamic networks. The underlying idea is to update the information in a rooted tree data structure that stores the edges of the network that are most relevant to the analysis. Extensive experiments on real-world and synthetic networks demonstrate that, with the exception of the inherently sequential component for creating the rooted tree, our proposed updatiing algorithm is scalable and, in most cases, also requires significantly less memory, energy, and time than recomputing-from-scratch algorithm. To the best of our knowledge, this is the first parallel algorithm for updating MST on weighted dynamic networks. The rooted-tree based framework that we propose in this paper can be extended for updating other weighted and unweighted tree-based properties such as single source shortest path and betweenness and closeness centrality.

Automatic extraction of sparse trees from high-resolution ortho-images

Obtaining information about tree species distribution in agricultural lands is a topic of interest for various applications, such as tree inventory, forest management, agricultural land management, crop estimation, etc. This information can be derived from images obtained from modern remote sensing technology, which is the most economical way as compare to field surveys covering large geographic areas. Therefore, in this study, a new method is proposed for extraction and counting of sparse and regular distributed individual pistachio trees from agricultural areas on large scale from high-resolution digital ortho-photo maps, which were obtained using an airborne sensor (Ultracam-X). The input images were first smoothed by applying Gaussian filter to reduce the impact of noise. Normalized difference vegetation indices (NDVI) were then derived to obtain vegetation areas followed by Otsu’s global thresholding algorithm to obtain candidate tree areas. Further, connected component (CC) analysis was applied to segregate each object. Morphological processing was performed to fill holes within tree objects and get smooth contours, which were obtained by using the Moore-neighbor tracing method (MNTM) for each CC, while geometrical constraints were applied to undermine possible non-tree elements from output image. To further improve the segmentation results for sparse trees, a new method was applied, called quadratic local analysis (QLA). QLA helped to segment the trees, which were missed by the Otsu method due to low contrast and resulted in improved accuracy (3–6%). The obtained results were compared with well-known support vector machine (SVM) classifier. Proposed method produced slightly better results (1–5%) than SVM for extraction of pistachio trees and obtained accuracy for QLA and SVM were 96 and 91% for region 1, while 91 and 90% for region 2 respectively.

Analysis of the genomic basis of functional diversity in dinoflagellates using a transcriptome-based sequence similarity network.

Dinoflagellates are one of the most abundant and functionally diverse groups of eukaryotes. Despite an overall scarcity of genomic information for dinoflagellates, constantly emerging high-throughput sequencing resources can be used to characterize and compare these organisms. We assembled de novo and processed 46 dinoflagellate transcriptomes and used a sequence similarity network (SSN) to compare the underlying genomic basis of functional features within the group. This approach constitutes the most comprehensive picture to date of the genomic potential of dinoflagellates. A core-predicted proteome composed of 252 connected components (CCs) of putative conserved protein domains (pCDs) was identified. Of these, 206 were novel and 16 lacked any functional annotation in public databases. Integration of functional information in our network analyses allowed investigation of pCDs specifically associated with functional traits. With respect to toxicity, sequences homologous to those of proteins found in species with toxicity potential (e.g., sxtA4 and sxtG) were not specific to known toxin-producing species. Although not fully specific to symbiosis, the most represented functions associated with proteins involved in the symbiotic trait were related to membrane processes and ion transport. Overall, our SSN approach led to identification of 45,207 and 90,794 specific and constitutive pCDs of, respectively, the toxic and symbiotic species represented in our analyses. Of these, 56% and 57%, respectively (i.e., 25,393 and 52,193 pCDs), completely lacked annotation in public databases. This stresses the extent of our lack of knowledge, while emphasizing the potential of SSNs to identify candidate pCDs for further functional genomic characterization.

Ballet E-learning using fuzzy set induced posture recognition by piece-wise linear approximation of connected components

The paper proposes a novel strategy of ballet e-learning for novices in remote areas. The present work is concerned with a training dataset of thirty four different ballet postures, obtained from images of trained dancers. A four-step procedure is primarily followed to extract features from the training images describing a specific posture p. First, the acquired images of K trained dancers of a specific posture p are pre-processed. The connected components (CCs) of the pre-processed images are then identified, followed by piece-wise linear approximation of their respective boundaries. After that, the internal angles of the polygonal boundary are used to characterize features of the given posture. This is repeated for all images of K trained dancers, all performing the given posture p. The merit of the paper lies in recasting the problem of selecting the CC boundary points for polygonal approximation as an optimization problem. The inter-subject uncertainty observed in the feature values extracted from the pre-processed images of a large number of experts/trained dancers, all executing the same posture p, here has been modeled by Gaussian fuzzy membership curve. The class of an unknown posture (obtained from the image of a novice) is determined by computing the support of each known posture class to the given unknown postural expression. The class with the maximum support (above a pre-defined threshold value) is declared as the conqueror. Experiments undertaken reveal that the accuracy of recognizing an unknown ballet posture of a novice, captured by a stereo camera, by the proposed stratagem is as high as 91.23%.

Multi-oriented text detection and verification in video frames and scene images

In this paper, we bring forth a novel approach of video text detection using Fourier-Laplacian filtering in the frequency domain that includes a verification technique using Hidden Markov Model (HMM). The proposed approach deals with the text region appearing not only in horizontal or vertical directions, but also in any other oblique or curved orientation in the image. Until now only a few methods have been proposed that look into curved text detection in video frames, wherein lies our novelty. In our approach, we first apply Fourier-Laplacian transform on the image followed by an ideal Laplacian-Gaussian filtering. Thereafter K-means clustering is employed to obtain the asserted text areas depending on a maximum difference map. Next, the obtained connected components (CC) are skeletonized to distinguish various text strings. Complex components are disintegrated into simpler ones according to a junction removal algorithm followed by a concatenation performed on possible combination of the disjoint skeletons to obtain the corresponding text area. Finally these text hypotheses are verified using HMM-based text/non-text classification system. False positives are thus eliminated giving us a robust text detection performance. We have tested our framework in multi-oriented text lines in four scripts, namely, English, Chinese, Devanagari and Bengali, in video frames and scene texts. The results obtained show that proposed approach surpasses existing methods on text detection.

An Adaptive Parallel Algorithm for Computing Connected Components

We present an efficient distributed memory parallel algorithm for computing connected components in undirected graphs based on Shiloach-Vishkin’s PRAM approach. We discuss multiple optimization techniques that reduce communication volume as well as load-balance the algorithm. We also note that the efficiency of the parallel graph connectivity algorithm depends on the underlying graph topology. Particularly for short diameter graph components, we observe that parallel Breadth First Search (BFS) method offers better performance. However, running parallel BFS is not efficient for computing large diameter components or large number of small components. To address this challenge, we employ a heuristic that allows the algorithm to quickly predict the type of the network by computing the degree distribution and follow the optimal hybrid route. Using large graphs with diverse topologies from domains including metagenomics, web crawl, social graph and road networks, we show that our hybrid implementation is efficient and scalable for each of the graph types. Our approach achieves a runtime of 215 seconds using 32 K cores of Cray XC30 for a metagenomic graph with over 50 billion edges. When compared against the previous state-of-the-art method, we see performance improvements up to 24 $\times$ .

Use of ResearchGate and Google CSE for author name disambiguation

Author name disambiguation plays a very important role in individual based bibliometric analysis and has suffered from lack of information. Therefore, some have tried to leverage external web sources to obtain additional evidence with success. However, the main problem is generally the high cost of extracting data from web pages due to their diverse designs. Considering this challenge, we employed ResearchGate (RG), a social network platform for scholars presenting their publication lists in a structured way. Even though the platform might be imperfect, it can be valuable when it is used along with traditional approaches for the purpose of confirmation. To this end, in our first (retrieval) stage we applied a graph based machine learning approach, connected components (CC) and formed clusters. Then, the data crawled from RG for the same authors were combined with the CC results in stage 2. We observed that 76.40% of the clusters formed by CC were confirmed by the RG data and they accounted for 68.33% of all citations. Second, a subset was drawn from the dataset by retaining those clusters having at least 10 members to examine the details. This time we additionally employed the Google Custom Search Engine (CSE) API to access authors' web pages as a complementary tool to RG. We observed an F score of 0.95 when CC results were confirmed by RG&CSE. Almost the same success was observed when only the CC approach was applied. In addition, we observed that the publications identified and confirmed through the external sources were cited to a greater extent than those publications not found in the related external sources. Even though promising, there are still issues with the use of external sources. We have seen that many authors present only a few selected papers on the web. This hampers our procedure, making it unable to obtain the entire publication list. Missing publications affect bibliometric analysis adversely since all citation data is required. That is, if only the data confirmed via external sources is used, bibliometric indicators will be overestimated. On the other hand, our suggested methodology can potentially decrease the manual work required for individual based bibliometric analysis. The procedure may also present more reliable results by confirming cluster members derived from unsupervised grouping methods. This approach might be especially beneficial for large datasets where extensive manual work would otherwise be required.

A Novel Feature Extraction Algorithm for IED Detection from 2-D Images using Minimum Connected Components

Buried Improvised Explosive Devices (IEDs) have become a significant threat to security forces combating terrorism. The detection of these concealed threats is a very challenging task. Ground Penetrating Radar (GPR) has shown promise in the detection of buried metallic and non-metallic IEDs or their components. The GPR produces a 2-D image of radar returns reflected off the buried objects. The challenge is how to detect IED’s in the presence of strong backscatter. In this paper, a Graph Theory based approach known as Minimum Connected Component (MCC) has been applied to detect buried objects from the 2-D images produced by the GPR. The MCC feature extraction algorithm efficiently extracted the IED component from ten different data sets collected by the GPR. The uniqueness of the algorithm is that it extracts the image of the IED without any user specified threshold or any user inputs.