Local Appearance Research Articles

Real-Time Object Tracking via Fusion of Global and Local Appearance Models

Real-Time Object Tracking via Fusion of Global and Local Appearance Models

A global FE–local analytical approach to modelling failure in localised buckles caused by crash

ABSTRACTLocalised buckles can occur in crashes when compressive stresses are dominant. However, due to their local appearance, they might be overlooked by a finite element analysis, meaning that local failure due to buckling will not be predicted. This could, for example, be catastrophic in the case of the crushing of a fuel tank or the impact on a ship hull, which could then lead to unpredicted and dangerous leakage. Methods to analytically assess the onset of such buckles are readily available. However, an off-the-shelf solution to predict the behaviour of them after the initial onset is not on hand. Therefore, an analytical method is developed that can predict the behaviour of a buckle after its onset, even in complex multi-axial states of stress that may be expected in crash scenarios. The strain over the buckle is captured as well, meaning that material failure can be predicted using existing material failure models. Also, the analysis is valid up to moderately thick shells, as would be expected in maritime structures. The proposed analytical method shows good agreement with a highly detailed finite element (FE) model, which only considers a single localised buckle. The goal would be to use the proposed method in conjunction with large-scale FE models, where the length of a shell element is equal to several times the thickness.

Automatic 3D liver segmentation based on deep learning and globally optimized surface evolution

The detection and delineation of the liver from abdominal 3D computed tomography (CT) images are fundamental tasks in computer-assisted liver surgery planning. However, automatic and accurate segmentation, especially liver detection, remains challenging due to complex backgrounds, ambiguous boundaries, heterogeneous appearances and highly varied shapes of the liver. To address these difficulties, we propose an automatic segmentation framework based on 3D convolutional neural network (CNN) and globally optimized surface evolution. First, a deep 3D CNN is trained to learn a subject-specific probability map of the liver, which gives the initial surface and acts as a shape prior in the following segmentation step. Then, both global and local appearance information from the prior segmentation are adaptively incorporated into a segmentation model, which is globally optimized in a surface evolution way. The proposed method has been validated on 42 CT images from the public Sliver07 database and local hospitals. On the Sliver07 online testing set, the proposed method can achieve an overall score of , yielding a mean Dice similarity coefficient of , and an average symmetric surface distance of mm. The quantitative validations and comparisons show that the proposed method is accurate and effective for clinical application.

An Empirical Study of Global Descriptors for Image-based Localization in Dense Urban Scenes

Background: Visual place recognition is an interesting technology that can be used in many domains such as localizing historical photos, autonomous navigation and augmented reality. The main stream of research in that domain was based on the use of local invariant features like SIFT. Little attention was given to region descriptors which can encompass local and global visual appearances. In this paper, we provide an empirical study on two main visual descriptors: (i) Local Binary Patterns, and (ii) Covariance matrices. Methods: In order to enhance the discriminative power of the final descriptor of each type, multi-block based descriptors are designed and compared. The descriptor corresponding to each input is formed by the concatenation of the features extracted from each building block. We show experimental results on matching test images with reference images acquired in dense urban scenes in the streets of the city of Paris. The problems of scale changes and occlusions are both treated by simulation. Different combinations of processing steps are treated (covariance and LBP descriptors, mono and multi-blocks, L1 and Chi-Squared distances). Results: The obtained results for the tested scenarios show an improvement in the classification rate for at least three scenarios when using multi-block based features rather than mono-block based ones. Conclusion: The use of multi-block based features can thus enhance the discrimination of the obtained final descriptor. The corresponding matching algorithms can lead to both high accuracy and scalability. Keywords: Global image descriptors, image-based localization, image retrieval, image matching.

Adaptive Combination of Local Motion, Appearance, and Shape for Video Segmentation

In this paper, we propose an accurate and robust video segmentation method. The main contributions are threefold: (1) multiple cues (appearance and shape) are explicitly used and adaptively combined to determine segment probability; (2) motion is implicitly used to compute the shape cue; and (3) the segment labeling is improved by utilizing geodesic graph cuts. Experimental results show the effectiveness of the proposed method.

SEISMIC FACIES ANALYSIS ON 2D SEISMIC REFLECTION PROFILE IN BARUNA AND JAYA LINE AT NORTH EAST JAVA BASIN

<p class="abstrak">Two dimension (2D) seismic profile of Baruna and Jaya lines at North-East Java Basin show seismic reflector characteristics that can be used to interpret sediment thickness and continuity. Those reflector characteristics that can be applied for seismic facies analysis that represent depositional environment. This study starts from seismic data processing that using Kirchhoff Post Stack Time Migration method which is 2D seismic profile as result. Seismic reflector characterization has been done to both 2D profiles. Seismic reflector characterization was grouped as (i) individual reflection, (ii) reflection configuration, (iii) reflection termination, (iv) external form. Individual reflection characteristics show high and medium amplitude, medium and low frequency, and continuous. Configuration reflection is continuous with parallel and subparallel type. Reflection termination shows onlap, and external form shows sheet drape. Local mound appearance can be interpreted as paleo-reef. Facies seismic anlysis result for this study area is shelf.</p>

Automatic segmentation of overlapping cervical smear cells based on local distinctive features and guided shape deformation

Automated segmentation of cells from cervical smears poses great challenge to biomedical image analysis because of the noisy and complex background, poor cytoplasmic contrast and the presence of fuzzy and overlapping cells. In this paper, we propose an automated segmentation method for the nucleus and cytoplasm in a cluster of cervical cells based on distinctive local features and guided sparse shape deformation. Our proposed approach is performed in two stages: segmentation of nuclei and cellular clusters, and segmentation of overlapping cytoplasm. In the first stage, a set of local discriminative shape and appearance cues of image superpixels is incorporated and classified by the Support Vector Machine (SVM) to segment the image into nuclei, cellular clusters, and background. In the second stage, a robust shape deformation framework is proposed, based on Sparse Coding (SC) theory and guided by representative shape features, to construct the cytoplasmic shape of each overlapping cell. Then, the obtained shape is refined by the Distance Regularized Level Set Evolution (DRLSE) model. We evaluated our approach using the ISBI 2014 challenge dataset, which has 135 synthetic cell images for a total of 810 cells. Our results show that our approach outperformed existing approaches in segmenting overlapping cells and obtaining accurate nuclear boundaries.

Fully automatic segmentation of femurs with medullary canal definition in high and in low resolution CT scans

Femur segmentation can be an important tool in orthopedic surgical planning. However, in order to overcome the need of an experienced user with extensive knowledge on the techniques, segmentation should be fully automatic. In this paper a new fully automatic femur segmentation method for CT images is presented. This method is also able to define automatically the medullary canal and performs well even in low resolution CT scans.Fully automatic femoral segmentation was performed adapting a template mesh of the femoral volume to medical images. In order to achieve this, an adaptation of the active shape model (ASM) technique based on the statistical shape model (SSM) and local appearance model (LAM) of the femur with a novel initialization method was used, to drive the template mesh deformation in order to fit the in-image femoral shape in a time effective approach.With the proposed method a 98% convergence rate was achieved. For high resolution CT images group the average error is less than 1mm. For the low resolution image group the results are also accurate and the average error is less than 1.5mm.The proposed segmentation pipeline is accurate, robust and completely user free. The method is robust to patient orientation, image artifacts and poorly defined edges. The results excelled even in CT images with a significant slice thickness, i.e., above 5mm. Medullary canal segmentation increases the geometric information that can be used in orthopedic surgical planning or in finite element analysis.

Dual-scale structural local sparse appearance model for robust object tracking

Recently, sparse representation has been applied in object tracking successfully. However, the existing sparse representation captures either the holistic features of the target or the local features of the target. In this paper, we propose a dual-scale structural local sparse appearance (DSLSA) model based on overlapped patches, which can capture the quasi-holistic features and the local features of the target simultaneously. This paper first proposes two-scales structural local sparse appearance models based on overlapped patches. The larger-scale model is used to capture the structural quasi-holistic feature of the target, and the smaller-scale model is used to capture the structural local features of the target. Then, we propose a new mechanism to associate these two scale models as a new dual-scale appearance model. Both qualitative and quantitative analyses on challenging benchmark image sequences indicate that the tracker with our DSLSA model performs favorably against several state-of-the-art trackers.

Robust visual tracking based on generative and discriminative model collaboration

Effective object appearance model is one of the key issues for the success of visual tracking. Since the appearance of a target and the environment changes dynamically, the majority of existed visual tracking algorithms tend to drift away from targets. To address this issue, we propose a robust tracking algorithm by integrating the generative and discriminative model. The object appearance model is made up of generative target model and a discriminative classifier. For the generative target model, we adopt the weighted structural local sparse appearance model combining patch based gray value and Histogram of Oriented Gradients feature as the patch dictionary. By sampling positives and negatives, alignment-pooling features are obtained based on the patch dictionary through local sparse coding, then we use support vector machine to train the discriminative classifier. The proposed method is embedded into a Bayesian inference framework for visual tracking. A combined matching method is adopted to improve the proposal distribution of the particle filter. Moreover, in order to adapt the situation change, the patch dictionary and discriminative classifier are updated by incremental learning every five frames. Experimental results on some publicly available benchmarks of video sequences demonstrate the accuracy and effectiveness of our tracker.

Comparative study on morpho-anatomy of leaf, stem and root of Boerhaavia diffusa L. (Nyctaginaceae) and its adulterant plants

Punarnava (Boerhaavia diffusa L.- Nyctaginaceae) is a promising drug to rejuvenate new cells in the body. It is well known in Ayurvedic medicine and locally called Tambadivasu. Superficially it is similar to other species of Boerhaavia and species of Trianthema and Sesuvium. Due to the minute morphological differences, the above plants are erroneously used in medicine as Punarnava, and at times on purpose as an adulterant. Therefore, it is necessary to highlight the anatomical features of Punarnava for proper identification of the medicinal plant species for local people and for scientific research. Due to the ambiguity in local names and similar apparent appearance, market samples of Punarnava are often adulterated with various species of Trianthema and Sesuvium. These adulterated samples contain neither the Punarnavine alkaloid, nor does it possess anisocytic stomata but possess paracytic stomata. Comparative study of stem anatomy showed two main characteristic differences. First, plenty of starch grains can be seen in both the ground parenchymatous tissues present in between successive cambia and xylem parenchyma of Punarnava which is not observed in species of Trianthema, and second, the phloem around the xylem of Punarnava root has semi-circular or eccentric patches, while that of Trianthema only has narrow strips. This study is focused on comparative SEM study of leaf morphologies and anatomy of leaf, stem, and root of Boerhaavia diffusa L., Trianthema portulacastrum L. and Sesuvium portulacastrum L.

Coupled-layer based visual tracking via adaptive kernelized correlation filters

Part-based visual model is particularly useful when the target appearance undergoes partial occlusion or deformation. The existing reliable patches tracking (RPT) method has achieved better result by identifying and exploiting the reliable patches that can be tracked correctly, yet it tends to fail in some challenging scenes since it ignores the holistic information of target completely, while, in fact, the target’s holistic appearance provides more discriminative features than local patches with low resolution. Based on the existing RPT and kernelized correlation filters tracking method, in this paper, we propose a coupled-layer visual model based tracker by combining the target’s global and local appearance in a coupled way. The global layer provides the holistic information and is treated as an approximation of the target. The local layer is composed of multiple small patches that are randomly initialized in the first frame. During tracking, the global tracker detects the target itself; its detection result is employed in the local layer to exploit the reliable patches and to estimate the target position corresponding to each patch. The exploited reliable patches are employed to estimate the target scale and to vote the current target location. Finally, both global and local models are updated with carefully designed updating mechanisms. Experiments conducted on 80 challenging benchmark sequences clearly show that our tracker improves the RPT tracker significantly both in overall and individual performance yet without obvious speed cost. Also, our tracker outperforms all the state-of-the-art trackers in overall datasets and eight independent datasets.

A Theoretical Approach on Face Recognition with Single Sample Per Class using CS-LBP and Gabor Magnitude and Phase

Objectives: To develop a theoretical model in order to understand how to recognize acquainted faces and the relationship between face processing acknowledgement with other aspects in face. Greater Performance is achieved in face recognition by local appearance based methods. Methods: The Centre Symmetric local binary pattern with gabor magnitude phase have been proposed in this paper to provide an expression, illumination and pose invariant in a single sample problem for face recognition approach with local spatial, scale and directional discriminate and low dimensional face representation based on features. The proposed methodology was compared with PCA and LBP. Findings: Gabor magnitude and Gabor phase tracks the texture boundaries of textured regions accurately. The evaluated Face features from CS-LBP and gabor magnitude phase has better performance. The Photometric descriptors are used in recent years, proven successful for computing regions which are in interest. In this approach the strength of SIFT descriptors are used in combination with LBP texture operator collectively called CS-LBP descriptor. This nullified illumination changes, strengthening flat image areas, and proficiency in computation. Improvements: For images with severe illumination variations SIFT descriptor is outperformed by CSLBP descriptor this was proved experimentally. The face recognition rate is increased by selective local texture feature Gabor Magnitude and Phase CS-LBP when compared with LBP method and Gabor filter.

Image Steganography Method to Improve the Stego Image Quality and Security using Reversible Texture Synthesis

Steganography is the profession of concealing a privacy message within a stated medium like image, audio or video. The main issue regarding every steganography technique is the changes that occur in the cover image after the data embedding procedure. We are using the source texture image for embedding privacy message in diversion with the cover image. This technique grants us to de-embedding the privacy message and source texture separating the stego synthetic texture. The advantages offering from this technique are embedding capacity that is symmetrical to the size of the stego texture image and the reversible potency to recover the source texture. The texture synthesis method is nothing but the generating a new large texture image with same local appearance and variable size from the small size image.

Local subspace smoothness alignment for constrained local model fitting

Constrained local model (CLM) is a classic method for facial landmarks estimation. While the CLM enhances the well-known active shape model with discriminative local appearance models, its shape model is based on the point distribution model, which is essentially principal component analysis over the training facial shape vectors and hence the nonlinear manifold of facial shapes is not well embedded. In this paper, we propose a novel manifold learning method, i.e., local subspace smoothness alignment (LSSA), to address this issue. The LSSA approach smoothes the nonlinear structure directly in the original feature space, with a newly defined geometric measure for the curvature of the local structures. We then proceed to apply this method for face alignment, with an ensemble of correlated local subspaces derived from LSSA. The proposed method is demonstrated on both toy data and real-world datasets that it yields reasonable manifold embedding and leads to encouraging performance for face alignment even under difficult conditions.

Visual Tracking via Coarse and Fine Structural Local Sparse Appearance Models.

Sparse representation has been successfully applied to visual tracking by finding the best candidate with a minimal reconstruction error using target templates. However, most sparse representation-based tracking methods only consider holistic rather than local appearance to discriminate between target and background regions, and hence may not perform well when target objects are heavily occluded. In this paper, we develop a simple yet robust tracking algorithm based on a coarse and fine structural local sparse appearance model. The proposed method exploits both partial and structural information of a target object based on sparse coding using the dictionary composed of patches from multiple target templates. The likelihood obtained by averaging and pooling operations exploits consistent appearance of object parts, thereby helping not only locate targets accurately but also handle partial occlusion. To update templates more accurately without introducing occluding regions, we introduce an occlusion detection scheme to account for pixels belonging to the target objects. The proposed method is evaluated on a large benchmark data set with three evaluation metrics. Experimental results demonstrate that the proposed tracking algorithm performs favorably against several state-of-the-art methods.

Retinal image registration via feature-guided Gaussian mixture model.

Registration of retinal images taken at different times, from different perspectives, or with different modalities is a critical prerequisite for the diagnoses and treatments of various eye diseases. This problem can be formulated as registration of two sets of sparse feature points extracted from the given images, and it is typically solved by first creating a set of putative correspondences and then removing the false matches as well as estimating the spatial transformation between the image pairs or solved by estimating the correspondence and transformation jointly involving an iteration process. However, the former strategy suffers from missing true correspondences, and the latter strategy does not make full use of local appearance information, which may be problematic for low-quality retinal images due to a lack of reliable features. In this paper, we propose a feature-guided Gaussian mixture model (GMM) to address these issues. We formulate point registration as the estimation of a feature-guided mixture of densities: A GMM is fitted to one point set, such that both the centers and local features of the Gaussian densities are constrained to coincide with the other point set. The problem is solved under a unified maximum-likelihood framework together with an iterative expectation-maximization algorithm initialized by the confident feature correspondences, where the image transformation is modeled by an affine function. Extensive experiments on various retinal images show the robustness of our approach, which consistently outperforms other state-of-the-art methods, especially when the data is badly degraded.

Shape estimation of subcutaneous adipose tissue using an articulated statistical shape model

The quantification of adipose tissue can give insights into obesity and the associated diseases. The segmentation of this tissue is however difficult, particularly due to the complex geometry and the local appearance similarities between the subcutaneous and visceral types. Shape priors can be used to regularise the segmentation of geometries with small variation in shape. However, human bodies are articulated and substantially different across subjects. In this paper, a novel method is proposed for the segmentation of the subcutaneous fat layer in full-body magnetic resonance imaging scans. The proposed method is based on a statistical shape model of the whole body surface, which is learned from geometric scans. The body model is factorised into pose and shape deformations, which allows a compact parametrisation of large variations in human shape. The proposed method is applied in the segmentation of five magnetic resonance imaging data-sets. The experiments show that the proposed model can be used to effectively segment the subcutaneous fat geometry in subjects with different body mass indices. The incorporation of the statistical model in the algorithm regularises the segmentation, and establishes correspondences between the subcutaneous fat layer of the geometries across subjects. The registration of the fat layer with a common geometry could facilitate the statistical analysis of the shape distribution across the different geometries.

The Development of the Hepatic Venous System in Man

The drainage of the umbilical and vitelline veins into the sinus venosus at 4 weeks of development (Carnegie Stage (CS) 12) is well documented, but to what extent these veins play a role in the development of the hepatic venous system remains to be established. We reinvestigated this issue in a closely spaced series of human embryos between 4 and 7 weeks of development. Embryological structures were visualised with Amira 3D reconstruction and Cinema 4D remodelling software.In the 5th week (CS13), the vitelline and umbilical veins became enclosed by the expanding mesenchyme of the septum transversum, while the expanding liver primordium enclosed the vitelline veins only. Concomitant with the repositioning of the entrance of the sinus venosus into the right atrium, the most distal part of the right vitelline vein (“hepatocardiac vein”) increased in diameter, whereas the corresponding left portion did not.In early CS14 embryos (~33 days of development), the liver primordium also enclosed the umbilical veins. While the connection of these veins with the sinus venosus disappeared, a connection (ductus venosus) appeared between the portion of the left umbilical vein that entered the liver and the right hepatocardiac vein. The intrahepatic part of the right vitelline vein temporarily persisted as a fairly large, “C”‐shaped vessel along the right‐sided outer edge of liver until the ductus venosus had acquired a similar size as the umbilical vein. Thereafter, the remaining part of the left and the intrahepatic portion of the right vitelline vein regressed. The right umbilical vein started to regress as soon as it was enclosed by the liver.In late CS14 embryos (~35 days of development), vessels larger than sinusoids emerged from the right vitelline (portal) vein and the (left‐sided) ductus venosus in the central part of the liver, while a well‐developed sinusoidal network was identifiable in the liver periphery. In CS15 and CS16 embryos, the portal branches extended further into the liver periphery, but remained confined to the right and left hemi‐livers, respectively. Meanwhile, the sinusoidal network extended towards the centre, followed by the appearance of hepatic veins in this network from the hepatocardiac vein towards the liver periphery. In CS16 embryos (~39 days of development), 2 main portal branches with only few side branches had developed in the right hemi‐liver, while the hepatic veins did not yet branch. In contrast, multiple small branches originated from the ductus venosus in the left hemi‐liver and hepatic veins were hardly identifiable. At 7 weeks of development (CS20), only 3 main portal branches remained in the left hemi‐liver and hepatic veins were appearing in between.ConclusionsThe vitelline and umbilical veins contribute to the hepatic venous system, but at their entrance and exit from the liver only. Development of the intrahepatic portal veins precedes that of the hepatic veins and development of the right hemi‐liver precedes that of the left hemi‐liver. The local appearance of hepatic veins does not obey the boundary between left and right hemi‐liver and appears to reflect the emergence of ordered flow patterns in that part of the liver. Our data explain the development and some of the variability of liver segments.

Combined discriminative global and generative local models for visual tracking

It is a challenging task to develop an effective visual tracking algorithm due to factors such as pose variation, rotation, and so on. Combined discriminative global and generative local appearance models are proposed to address this problem. Specifically, we develop a compact global object representation by extracting the low-frequency coefficients of the color and texture of the object based on two-dimensional discrete cosine transform. Then, with the global appearance representation, we learn a discriminative metric classifier in an online fashion to differentiate the target object from its background, which is very important to robustly indicate the changes in appearance. Second, we develop a new generative local model that exploits the scale invariant feature transform and its spatial geometric information. To make use of the advantages of the global discriminative model and the generative local model, we incorporate them into Bayesian inference framework. In this framework, the complementary models help the tracker locate the target more accurately. Furthermore, we use different mechanisms to update global and local templates to capture appearance changes. The experimental results demonstrate that the proposed approach performs favorably against state-of-the-art methods in terms of accuracy.