MPEG-7 Dataset Research Articles

An algorithm for extracting similar segments of moving target trajectories based on shape matching

Trajectory similarity analysis of moving target is the foundation for mining high-value and regular behavioral information such as motion preferences, activity hotspots and frequent paths. Unlike most trajectory similarity analysis methods aimed at discovering correlations of target activities in time, space or spatio-temporal domains, this paper focuses on the shape matching of target trajectories. If some specific shapes frequently appear in historical trajectories, extracting these local shapes would be beneficial for analyzing the target motion templates and behavior modes. Trajectory segments with similar shapes may not have spatio-temporal correlation, and the shapes also have geometric transformation characteristics such as rotation, scaling and translation. Since the existing trajectory similarity analysis methods cannot be directly applied, an algorithm for extracting similar segments based on shape matching is proposed. First, a new shape descriptor based on signed barycenter distance (SBD) is established. It describes a trajectory as a one-dimensional shape feature sequence, which has the advantage of low computational complexity. Then, the distributed nearest neighbor search strategy is used in the particle swarm optimization (PSO) method, which aims to accelerate the retrieval of trajectory segments with similar shapes and improve the matching accuracy. Experiments on MPEG-7, handwritten character and maneuvering target simulation trajectory data sets show that compared with the existing typical shape descriptors, SBD shape descriptor has advantages in accuracy and noise insensitivity, and the improved PSO method can efficiently and accurately obtain the local shape matching results.

Simultaneous Boundary and Interior Parameterization of Planar Domains Via Deep Learning

In isogeometric analysis (IGA), domain parameterization plays a crucial role as it significantly impacts the results of subsequent numerical analyses. Previous literature has often treated domain parameterization as two separate processes: boundary correspondence and interior parameterization. However, this division tends to decrease the quality of the final parameterization. To address this issue, this article proposes a learning-based method that simultaneously parameterizes the boundary and interior of a simply connected planar computational domain using a deep neural network. In our approach, we train a neural network to represent the inverse parameterization, which maps the computational domain to the parametric domain, based on the given boundary curve of the planar computational domain. The key component of our neural network is a Multilayer Perceptron (MLP), which takes the coordinates of a point in the computational domain as input and outputs the corresponding parametric values. The loss function of our method comprises three terms. First, it considers the area distortion and angular distortion of the parameterization, which are determined by the Jacobian matrix of the parameterization. Second, it incorporates the boundary difference, quantified by the Hausdorff distance between the boundary of the parametric domain and the image of the inverse mapping of the computational boundary. By optimizing this loss function, we achieve a robust parameterization. Once the network is trained, we employ a tensor-product B-spline function to fit the mapping from the parametric domain to the computational domain. To evaluate the effectiveness of our method, we conduct experiments on the MPEG-7 dataset. The experimental results demonstrate that our approach yields parameterization results with lower distortion and higher bijectivity ratio compared to state-of-the-art approaches.

Multi-level contour combination features for shape recognition

We present a novel multi-level contour combination feature for shape recognition. This combination feature effectively solves large intra-class changes and nonlinear deformations of object shapes, thereby enhancing the performance of shape recognition. First, we divide the shape contour into two levels: the sampling points and the contour fragments, where sampling points are used to describe the detailed information of a shape and contour fragments are used to represent the global feature of a shape. Second, we employ the Fisher vector (FV) approach to encode the local sampling point feature and contour fragment feature as high-level characteristics. Finally, we combine the high-level characteristics after FV encoding and perform shape recognition through a linear support vector machine (SVM) model. The proposed method has been assessed on three benchmark shape datasets, including the Animal, MPEG-7,and ETH-80 datasets. Our method achieves 92.70%, 99.26% and 98.32% classification accuracy on the Animal, MPEG-7, and ETH-80 datasets, respectively. In addition, our method can also be applied to the classification of objects in real-word scenes. We combine the Weizmann Horse and the ETHZ Cow real-world scene datasets, and our method achieves 99.25% classification accuracy on the combined dataset. The recognition results of our approach are better than prior state-of-the-art shape recognition methods, which demonstrate the effectiveness and superiority of our approach.

L-shaped geometry-based pattern descriptor serving shape retrieval

Feature representation patterns serving shape retrieval have gained considerable attention over recent years. Accordingly, a geometry-based characterization arrangement based on the l-shape pattern is adopted in this paper for shape description. The presented l-shape pattern descriptor bounds shape edges for providing highly localized distinct features supporting characterization. Then a novel feature representation scheme fabricates these shapes into histograms subsequently, employed for matching and retrieval. The metric Bull’s Eye Retrieval (BER) rate is deployed for retrieval analysis on the Kimia-99, MPEG-7 and Tari-1000 datasets that reveal a uniform and remarkable performance higher than 90% over its predecessors. The congruence nature of the l-shaped geometrical arrangement ensures its robustness towards diverse affine transformations and warrants increased performance. The complexity associated with descriptor realization along space and time reveals its lightweight simplicity and efficiency.

Quadrant-based contour features for accelerated shape retrieval system

Abstract Shape representation and retrieval are essential research topics of computer vision. This paper proposes a novel feature set to be used in content-based image retrieval systems. The proposed method is an extended version of our previous study which uses contour information of shapes. The previous study calculated the center of mass (CoM) of the shape. By taking the CoM as origin, we created imaginary vectors in every angular direction. From each vector, we extracted three features which are the number of intersections between vector and contour, average distance of intersection points to CoM, and standard deviation of these points. In this method, we extract novel features and decrease the size of the feature set to decrease the computation time. We divide the shape into quadrants and represent each quadrant by nine features. Each shape image is represented by a 4x9 feature vector. We tested the proposed method on MPEG-7 and ETH-80 datasets and compared it with the state-of-art. According to the results, our method decreased the computation time dramatically while giving a state-of-art level retrieval accuracy.

A Fast and Efficient Shape Blending by Stable and Analytically Invertible Finite Descriptors.

In a previous work, we have proposed a morphing method based on invertible and stable descriptors that are invariant to Euclidean transformations and to the starting point. The stability guarantees the closeness in the shape sense of the reconstructed intermediate contours. However, this set of descriptors is not defined by a general expression. Here, we propose several sets of stable and invertible finite descriptors expressed by the same formula. Its stability is proven for a subset of this family thanks to the finite-dimension of the invariant space. Such finite dimension results from the Discrete Fourier Transform model that is used instead of Fourier coefficients. Moreover, an analytical general inverse formula is established. The use of the inverse analytical formula and the double utilization of the Fast Fourier Transform ensure an effective blending while being computationally efficient. Finally, we propose a new quantitative criterion for shape morphing. The latter is based on the Euclidean distances between successive curves in a morphing sequence after applying a given registration. Therefore, it allows us to compare the blending results of each set of descriptors. Several experiments are conducted on KIMIA'99 and MPEG-7 datasets. The results highlight the concordance of this criterion with the morphing visual quality and indicate which set of descriptors generates the most appropriate blending.

Efficient retrieval of trademark images from large databases

Trademarks law is considered as the most pervasive amongst all the intellectual property laws because all the judgments require contemplating the imaginations of the consumers. Once the trademark has been submitted for registration, the examiners at the trademark office make sure that it is not similar to any of the previous registered trademarks. This motivated the need of an automated trademark retrieval system. This paper makes a contribution in the field of trademark image retrieval by proposing a retrieval technique that allows a flexible combination of colour, texture and shape features. Moreover the proposed technique utilises HSV colour histogram, for colour, multi resolution Gabor wavelet for texture and an integration of Zernike moments for global shape and scale invariant feature transform (SIFT) for local shape feature extraction. The results have been tested on MPEG7, MPEG trademark, WANG and self-compiled datasets. The improvement achieved in precision is 14% on MPEG7, 30% on the WANG and 38% on self-compiled dataset. Similarly, 26% improvement in average recall is achieved and around 16% when our proposed shape feature is compared with other state of the art techniques like Fourier descriptors, Hu moments, wavelet descriptors, Zernike moments and edge gradient co-occurrence matrix.

Fast global SA(2,R) shape registration based on invertible invariant descriptor

Here, we intend to introduce a fast and non-rigid global registration for simple planar closed curves relatively to the planar Special Affine group SA(2,R). In previous work, Ghorbel (1998) has been introduced a complete and stable set of invariant for simple closed planar curves which is invariant jointly under its original parametrization and special planar affine transformations. Such property allows us a robust reconstruction of the considered object up to a special affine transformation. In this paper, several numerical difficulties of the computation of the proposed reconstruction are considered. The robustness of this inverse problem with respect to noise and reasonable deformations of non-rigid shape is demonstrated experimentally. The proposed new registration is based, on the one hand, on the shift theorem relating to the group SA(2,R) and, on the other hand, on the invertibility of the set of invariant. Since this shift theorem allows the extraction of a pose parameters exciting between reference and target objects. The low algorithmic complexity is due to the fact that the computation of the inverse descriptors are based essentially on the Fast Fourier Transformation (FFT) algorithm. Experiments are conducted on different known datasets such as MPEG-7, MCD, Kimia-99, Kimia216, ETH-80 and Swedish leaf datasets. Promising results on the sense of shape retrieval and shape recognition rates will also be demonstrated.

Fast Plant Leaf Recognition Using Improved Multiscale Triangle Representation and KNN for Optimization

Due to the complexity and similarity of plant leaves, it is very important to study an effective leaf-feature extraction method to improve the recognition rate of plant leaves. We study five multi-scale triangle representations: the triangle unsigned area representation (TUA), the triangle vertex angle representation (TVA) and three new representations, which we define as the gray average (TGA), the gray standard deviation (TGSD) and the side length integral (TSLI) on the triangle. In this method the curvature features of the contour, the texture features and the shape area feature are extracted to provide a multiscale leaf-feature description, and a new adaptive KNN for optimization method is proposed to improve the retrieval rate of leaf datasets. Experiments show that compared with the state-of-the-art methods, our method has higher accuracy on the Swedish and Flavia plant leaf datasets, which are respectively 99.35% and 99.43% with 84.76% Mean Average Precision (MAP) value and has comparable results on MPEG-7, kimia99 and kimia216 datasets. When our method is combined with KNN for optimization, the retrieval rate of the above datasets has been significantly improved, especially MAP on the Flavia dataset increases to 94.48%.

2D Shape Recognition and Retrieval Using Shape Contour Based on the 8-Neighborhood Patterns Matching Technique

A technique for 2D shape recognition and retrieval is proposed. The proposed technique is based on the 8-neighborhood pattern which represents each point or pixel on the contour of the shape. These patterns are used as a framework in matching the shape of the object. The recognition and retrieval process are conducted by traversing through the contour of the shape and analyzes each point on the contour by considering the 8-neighborhood pattern. The 8-neighborhood patterns are assigned unique labels which are computed on their every occurrence during contour traversal. The cost of the best match between the shapes is evaluated by comparing the hit value obtained by the contour traversal of the shapes to be matched. The recognition and retrieval are carried out using the leave-one-out strategy and standard bull eye score, respectively. The proposed method is experimented on the MPEG-7 data set and the chicken piece data set. The results both for recognition and retrieval outperform most of the previously proposed methods.

SHAPE DESCRIPTOR FOR OBJECT CLASSIFICATION

In this article we propose an efficient modification of a previously published shape descriptor, which is fast and simple to compute. Proposed descriptor was developed for object classification and should be used with classifiers: SVM, KNN, etc. Object classification is a common task of computer vision, which has many applications in different areas: computer intelligence, robotic vision, smart cameras, autonomous driving, etc. Because the properties of objects are largely determined by their geometric features, shape analysis and classification are essential to almost every applied scientific and technological area. Main steps of the proposed algorithm are as follows: to find the object bounds; to smooth the bound contour using extremes based approach (if needed); to find side contour feature for each of N object rotation; to gather common feature vector; to classify object contour, using pre-trained classifier (SVM, KNN). In this work, in addition to method modifications (saving object proportions, rotation invariance, applying KNN classifier), we provide a wide comparison of our algorithm with existing approaches. The described method provided state-of-the-art performance on 100 leaves and Mpeg7 datasets, and showed good results on our own Mushroom dataset separately or together with texture or color based features.

Pattern recognition based on compound complex shape-invariant Radon transform

Radon transform based on complex shape detection is one of the most important challenges in the field of pattern recognition. Finding a robust transform for extracting and identifying complex shape is still an ongoing research issue. In this paper, aiming to simplify complex pattern recognition, we propose a new methodology of recognition flows. The proposed method is based on a decomposition of complex objects into elementary shapes and an evaluation of the pertinence of the generated primitives in recognizing the object. Accordingly, a series of the most pertinent primitives are selected. Then, the recognition step starts with a test on the most pertinent primitives. Our approach, called compound complex shape-invariant Radon transform, takes its importance in recognizing object class from a partial elementary scale and orientation-invariant Radon transforms, done on selected primitives. SVM classifier is then used for decision making based on obtained transforms. Validation of the proposed approach is done on the MPEG7 dataset, yielding a recognition rate of 98%.

Tetrakis square tiling-based triangulated feature descriptor aiding shape retrieval

Ongoing research addressing shape retrieval render highly acute feature descriptors that are computationally intensive. Hence, a simple approach through a novel tessellated version of the Tetrakis square tiling for acute feature characterization aiding shape retrieval is contributed in this paper. The proposed descriptor, named Triangulated Feature Descriptor (TFD), performs feature characterization and abstraction by fusing hybrid geometrical concepts. The mechanism initially tiles the image into square regions that are later tessellated into right-angled triangles. The right-angled triangular neighbors interact locally using the trigonometric identities to produce an angle-based feature map representing an image. The attained feature maps are locally segmented to finally produce a shape histogram. Since, the intended descriptor is modeled based on intrinsic angular interactions between the local neighbors the resulting descriptor is highly invariant to diverse affine transformations. This characteristic is examined by performing rigorous experiments on MPEG-7, TARI-1000, Articulated and PHOS datasets. Competitive analysis reveals a consistent retrieval rate greater than 85% achieved by TFD with increased performance scores.

Shape automatic clustering-based multi-objective optimization with decomposition

In this paper, a new shape automatic clustering method based on multi-objective optimization with decomposition (MOEA/D-SAC) is proposed, which aims to find the final cluster number k as well as an optimal clustering result for the shape datasets. Firstly, an improved shape descriptor based on the shape context is proposed to measure the distance between shapes. Secondly, the diffusion process is applied to transform the similarity distance matrix among total shapes of a dataset into a weighted graph, where the shapes and their distance are regarded as nodes and weight of edges, respectively. Thirdly, a new clustering method called “the soft clustering” is devised, starting with constructing an adjacency graph which can maintain the edges with the weights of k-nearest-neighbor nodes. Then, a multi-objective evolutionary algorithm with decomposition (MOEA/D) is applied to achieve automatic graph clustering scheme. The proposed clustering algorithm has been used to cluster several shape datasets, including four kimia datasets and a well-known MPEG-7 dataset, and experimental results show that the proposed method can demonstrate competitive clustering results.

Deformation Based Curved Shape Representation.

In this paper, we introduce a deformation based representation space for curved shapes in . Given an ordered set of points sampled from a curved shape, the proposed method represents the set as an element of a finite dimensional matrix Lie group. Variation due to scale and location are filtered in a preprocessing stage, while shapes that vary only in rotation are identified by an equivalence relationship. The use of a finite dimensional matrix Lie group leads to a similarity metric with an explicit geodesic solution. Subsequently, we discuss some of the properties of the metric and its relationship with a deformation by least action. Furthermore, invariance to reparametrization or estimation of point correspondence between shapes is formulated as an estimation of sampling function. Thereafter, two possible approaches are presented to solve the point correspondence estimation problem. Finally, we propose an adaptation of k-means clustering for shape analysis in the proposed representation space. Experimental results show that the proposed representation is robust to uninformative cues, e.g., local shape perturbation and displacement. In comparison to state of the art methods, it achieves a high precision on the Swedish and the Flavia leaf datasets and a comparable result on MPEG-7, Kimia99 and Kimia216 datasets.

A string grammar fuzzy-possibilistic C-medians

In some pattern recognition problems, the syntactic or structural information that describes each pattern is important. A syntactic pattern can be described using string grammar. There is only a handful of research works involving with the string grammar clustering. The string grammar hard C-means (sgHCM) is one of the most well-known clustering algorithms in syntactic pattern recognition. Since, it has been proved that fuzzy clustering is better than a hard clustering, a string grammar fuzzy C-medians (sgFCMed) algorithm to improve the sgHCM was previously proposed. However, the sgFCMed may not provide a good clustering result for any application with an overlapping data. Thus, in this paper, a string grammar fuzzy-possibilistic C-medians (sgFPCMed) algorithm is introduced to cope with the overlapping data problem. The proposed algorithm is implemented on four real overlapping data sets, i.e., MPEG-7 data set, Copenhagen chromosomes data set, MNIST database of handwritten digits, and USPS database of handwritten digits. The proposed sgFPCMed results are compared with the results from the sgHCM and the sgFCMed. The results show that the proposed sgFPCMed is better than both. The proposed sgFPCMed algorithm results are directly and indirectly compared with the results from other syntactic or numeric methods The proposed sgFPCMed is better than some approaches and comparable to some of the methods However, since the proposed sgFPCMed is a string grammar clustering, it is easier to transform each prototype string back into the original form of data set.

Landmark-Constrained Elastic Shape Analysis of Planar Curves

ABSTRACT Various approaches to statistical shape analysis exist in current literature. They mainly differ in the representations, metrics, and/or methods for alignment of shapes. One such approach is based on landmarks, that is, mathematically or structurally meaningful points, which ignores the remaining outline information. Elastic shape analysis, a more recent approach, attempts to fix this by using a special functional representation of the parametrically defined outline to perform shape registration, and subsequent statistical analyses. However, the lack of landmark identification can lead to unnatural alignment, particularly in biological and medical applications, where certain features are crucial to shape structure, comparison, and modeling. The main contribution of this work is the definition of a joint landmark-constrained elastic statistical shape analysis framework. We treat landmark points as constraints in the full shape analysis process. Thus, we inherit benefits of both methods: the landmarks help disambiguate shape alignment when the fully automatic elastic shape analysis framework produces unsatisfactory solutions. We provide standard statistical tools on the landmark-constrained shape space including mean and covariance calculation, classification, clustering, and tangent principal component analysis (PCA). We demonstrate the benefits of the proposed framework on complex shapes from the MPEG-7 dataset and two real data examples: mice T2 vertebrae and Hawaiian Drosophila fly wings. Supplementary materials for this article are available online.

TSS & TSB: Tensor scale descriptors within circular sectors for fast shape retrieval

We propose two novel region-based descriptors for shape-based image retrieval and analysis, which are built upon an extended tensor scale based on the Euclidean Distance Transform (EDT). First the tensor scale algorithm is applied to extract local structure thickness, orientation, and anisotropy as represented by the largest ellipse within a homogeneous region centered at each image pixel. In this work, we extend the local orientation to 360°. Then, for the first proposed descriptor, named Tensor Scale Sector descriptor (TSS), the local distributions of relative orientations within circular sectors are used to compose a fixed-length feature vector for a region-based representation. For the second method, named Tensor Scale Band descriptor (TSB), we consider histograms of relative orientations for each circular concentric band to compose a fixed-length feature vector with linear time matching. Experimental results with MPEG-7 and MNIST datasets are presented to illustrate and validate the methods. TSS can achieve high retrieval values comparable to state-of-the-art methods, which usually rely on time-consuming correspondence optimization algorithms, but uses a simpler and faster distance function, while the even faster linear complexity of TSB leads to a suitable and better solution for very large shape collections.

A Study of Sub-Pattern Approach in 2D Shape Recognition Using the PCA and Ridgelet PCA

In the area of computer vision and machine intelligence, image recognition is a prominent field. There have been several approaches in use for 2D shape recognition using shape features extraction. This paper suggest, subspace method approach. Normally in the earlier methods proposed so far, an entire image is considered in the training and matching operation, with sub pattern approach a given image is partitioned in to many sub images. The recognition process is carried out in two steps, in the first step the Ridgelet transform is used to feature extraction, in the second step PCA is used for dimensionality reduction. For recognition efficiency rate a test study is conducted by using seventeen different distance measure technique. The training and testing process is conducted using leave-one-out strategy. The proposed method is tested on the standard MPEG-7 dataset. The results of Ridgelet PCA are compared with PCA results.

Sparse Contextual Activation for Efficient Visual Re-Ranking.

In this paper, we propose an extremely efficient algorithm for visual re-ranking. By considering the original pairwise distance in the contextual space, we develop a feature vector called sparse contextual activation (SCA) that encodes the local distribution of an image. Hence, re-ranking task can be simply accomplished by vector comparison under the generalized Jaccard metric, which has its theoretical meaning in the fuzzy set theory. In order to improve the time efficiency of re-ranking procedure, inverted index is successfully introduced to speed up the computation of generalized Jaccard metric. As a result, the average time cost of re-ranking for a certain query can be controlled within 1 ms. Furthermore, inspired by query expansion, we also develop an additional method called local consistency enhancement on the proposed SCA to improve the retrieval performance in an unsupervised manner. On the other hand, the retrieval performance using a single feature may not be satisfactory enough, which inspires us to fuse multiple complementary features for accurate retrieval. Based on SCA, a robust feature fusion algorithm is exploited that also preserves the characteristic of high time efficiency. We assess our proposed method in various visual re-ranking tasks. Experimental results on Princeton shape benchmark (3D object), WM-SRHEC07 (3D competition), YAEL data set B (face), MPEG-7 data set (shape), and Ukbench data set (image) manifest the effectiveness and efficiency of SCA.