最新刊期
卷 16 ， 期 8 ， 2011
- 摘要：Sub-pixel mapping of remotely sensed imagery is a post-processing method of pixel unmixing in order to solve the issue of spatial uncertainty of mixed pixels. With the fraction images derived from pixel unmixing technologies as input, the spatial distribution of each land cover class within mixed pixels is determined and a sub-pixel scale land cover map is generated according to the spatial correlation principle. This paper firstly introduces the concept of sub-pixel mapping and analyses the theoretial models and currently used algorithms. Then, the main sources of errors, different accuracy assessments of the model and the ways to present the uncertainty of the model are summarized. The main methods for increasing the accuracy of sub-pixel mapping by ancillary data are also discussed. Finally, this paper concludes with an outlook and several key techniques of sub-pixel mapping.关键词：remote sensing;mixed pixels;sub-pixel mapping;super-resolution mapping;spatial distribution;scale4140|341|0更新时间：2024-05-07
- 摘要：The key point of the global motion estimation is how to select a global motion model. In this paper, integrating with a rate distortion optimization algorithm, an adaptive global motion estimation method to optimize the coding is proposed. The scheme of the algorithm is as follows: for the same image, using the translated motion model, affine motion model and quadratic motion model to estimate the inter frame motion and calculating their Lagrangian costs respectively. The motion model which yields least Lagrangian cost is selected as the best motion mode for current frame coding. Simulated results show that the proposed technique is robust and have better performance for various resolutions of video sequences.关键词：global motion model;global motion estimation;rate distortion optimization2793|221|0更新时间：2024-05-07
- 摘要：The extraction of SPN (sensor pattern noise) is a key procedure for images source identification. But the SPN extracted by existing methods are susceptible to the interference of texture and scene in images. For these reasons, a novel method is proposed for SPN extraction based on combined filtering. Firstly, images are analyzed with an orthogonal wavelet transform, an edge preserving bilateral filter and a local adaptive minimum mean squared error filter for the approximation and detail subband respectively. Then we denoised with bilateral filter in spatial domain. Finally, the classifier based on correlation detection principle is constructed with 9 cameras reference SPN, and accurate identification for SPN has been achieved. Then we analyzed the results of 3 kinds of typical size. The experimental results indicate that the proposed method effectively reduces the interference due to scene. A 79.32% accuracy has been achieved even the size of detected images are 256×256 pixels.关键词：image source;non intrusive identification;sensor pattern noise;correlation;filtering2679|215|0更新时间：2024-05-07
- 摘要：The degradation of foggy images is non linear with the depth of field, a new adaptive Retinex algorithm in foggy image enhancement is proposed. The algorithm adjusts the parameters of Retinex automatically according to the strong degree of fog reflected by the features of cumulative distribution function. A new calculation model of image enhancement is built based on this. The model combines the results enhanced by adaptive Retinex algorithm and CLAHE algorithm with different weights adaptively, makes enhanced images be able to keep both color constancy and luminance constancy, and make foggy images clear. Subjective observation and objective evaluation show that, the proposed method is more effective than HE and MSR algorithms both in details enhancement and color preserving.关键词：foggy image enhancement;Retinex algorithm;CLAHE algorithm;color constancy;illumination constancy2989|352|0更新时间：2024-05-07
- 摘要：H.264/AVC is the newest video compression coding standard. The intra prediction for H.264 uses RDO to choose the best prediction mode. This makes the coding performance greatly enhanced. But it also causes the coding complexity and the computation obviously to increase. This paper studies several fast intra prediction algorithms, and proposes a novel fast intra prediction algorithm which fuses the macroblocks smooth character and the 4×4 blocks texture property. The algorithm pre decides the size of block by macroblocks smooth character and decreases the predict modes by the texture character of 4×4 block. The result compared with JM95, shows that the algorithm makes the coding complexity reduced 71.3% while the PSNR is almost invariable.关键词：H.264;intra prediction;RDO;prediction modes2636|200|0更新时间：2024-05-07
- 摘要：A restriction item that is a nonlinear heat equation is attached to a variational IAC (integrated active contour) model on the basis of analysis on regions and edges information from all channels of the valued vector image, which forces level set to maintain the signed distance function properties without the costly re initialization. A balance function for diffusion rate of the nonlinear heat equation is introduced into this model, and therefore the level set evolution segmentation process becomes fast and stable. In addition, an efficient discretization method with spatial rotation invariance gradient and divergence operator are proposed as numerical implementation scheme. Finally, the experiments on some images have demonstrated the efficiency, accuracy and robustness of the proposed method.关键词：image segmentation;IAC model;valued vector image;divergence operator3028|252|0更新时间：2024-05-07
- 摘要：Firstly, an image segmentation method based on level set evolution without re initialization is studied. This method depends on edge indicator function. However, the method is inaccurate in capturing the sharp corners of the object during image segmention. Then, we analyze the method for image segmentation by using the edge indicator function's curvature information to identify the sharp convex corners. However, the model can only capture the sharp convex corners of the object. When the object has sharp concave corners, the model fails to capture the ones accurately. To overcome this difficulty, we employ the image's curvature information to identify the sharp concave corners, and then combine it with the method of identifying the sharp convex corners by using the edge indicator function's curvature information. By further modifying the edge indicator function to make the method capture both the sharp convex corners and the sharp concave corners of the object more accurately. The numerical experiments show the advantage of the improved model.关键词：image segmentation;level set method;edge indicator function;without-re initialization;sharp corners.2683|265|0更新时间：2024-05-07
- 摘要：The cross entropy in the existing thresholding methods does not satisfy the symmetricity of distance measure. And the computation speed of the algorithms can be further improved. Thus an image threshold selection method based on decomposition and two dimensional symmetric cross entropy is proposed in this paper. Firstly, the difference between the segmented image and the original one is measured by the symmetric cross entropy. The threshold selection formulae are derived based on the one dimensional and two dimensional symmetric cross entropy, respectively. A two dimensional fast recursive algorithm is given, which makes the computation complexity reduced to O(L2) from O(L4) of full search. Then the computation of two dimensional symmetric cross entropy is converted into two one dimensional spaces and its computation complexity is further reduced to O(L). The experimental results show that, compared with the existing threshold selection method based on two dimensional nonsymmetric cross entropy, the proposed method has stronger anti noise and the processing time is significantly reduced. It is an effective threshold selection method based on two dimensional cross entropy.关键词：image segmentation;threshold selection;symmetric cross entropy;two dimensional histogram;recursive algorithm;decomposition3250|345|0更新时间：2024-05-07
- 摘要：The detection of solitary pulmonary nodules (SPN) is proven to be of critical importance in early stage lung cancer diagnosis. Aiming at reducing the false positive regions caused by the dot enhancement filter which is sensitive to the lung nodules, a new recognition method based on calculation of three dimensional (3D) enhancement density index and decision rule is proposed. An adaptive bilateral filter is applied to reduce the noisy and smooth CT image sequences. Then, the pre enhancement coefficients and volume of interest (VOI) are obtained by computing the Hessian matrix and corresponding eigenvalues. After analyzing the distribution of pre enhancement coefficients, 3D enhancement density index is constructed. Finally, a decision rule is adopted to identify nodule candidates. The proposed method is tested on two lung CT image sets. The experimental results illustrate the efficiency of the proposed algorithm.关键词：dot enhancement;adaptive bilateral filter;Hessian matrix;enhancement density index3071|218|0更新时间：2024-05-07
- 摘要：The positive fuzzy rules often were used only for image classification in the traditional image classification system, while the negative image classification rules were ignored in effect. Nguyen introduced the negative Fuzzy rules into the image classification, proposed a combination of positive and negative fuzzy rules to form the positive and negative fuzzy rule system, and then applied it to remote sensing image/natural image classification. Their experiments proved that their proposed method has achieved good results. However, since their method was realized using the feed forward neural network model which adjust the weights in the gradient descent, the training speed is very slow. Extreme learning machine (ELM) is a single hidden layer feed forward neural network (SLFN) learning algorithm, which has advantages such as quick learning, good generalization performance. In this paper,it proves that Extreme Learning Machine (ELM) and the positive and negative fuzzy rule system is essentially equivalent, so ELM can be naturally used for image classification. Our experimental results support this claim.关键词：image classification;positive and negative fuzzy rules;extreme learning machine3042|262|0更新时间：2024-05-07
- 摘要：To solve the road oriented marking recognition problem in complex driving environment, this paper presents a novel approach with two steps: 1) merge multi scale autoconvolution (MSA) moment with affine invariant and image region property to define a new feature descriptor to describe road oriented marking; 2) apply support vector machine (SVM) to classify the image objects. The experimental data demonstrate that the new descriptor has an approximate projective invariance and is suitable for road oriented marking classification in different visual angles and partly being occluded .关键词：road oriented marking;multi scale autoconvolution (MSA);affine invariant moment;support vector machine2922|169|0更新时间：2024-05-07
- 摘要：To consider the differences between infrared image and common gray level image, the effects of regional division are not good when the commonly used gray level-average gray level 2D histogram is used in infrared image segmentation. An improved Otsu threshold selection method based on gray level-gradient 2D histogram is proposed in this paper. The appropriate gradient operator is chosen. An improved particle swarm optimization(PSO) algorithm is used to search for the segmentation threshold. An effective method that identifies premature stagnation is embedded to PSO, so once premature stagnation happens, a randomized solution, as a substitute for current optimum, is used for particles to go out of the local optima. The simulation experiments demonstrate that the algorithm proposed in this paper achieves accurate borders and clear details of features after infrared image threshold because of the new 2D histogram. The compute speed is improved effectively.关键词：infrared image segmentation;threshold selection;gradient;2D-Otsu method;particle swarm optimization(PSO)3208|213|0更新时间：2024-05-07
- 摘要：Face detection has wide applications. But many of the detection algorithms are sensitive to illumination variation. Detection rates usually drop quickly if the illumination is too bright, too dark, or non-uniform illumination. This paper proposed a adaptive logarithm illumination compensation method based neighbour remediation technique, the method use logarithm function as primary function, therefore it's true of human visual trait; and setup the offset's coefficients of logarithm function to solve the illumination problem in varying degree, thus this method is ability of compensation of all types of image in varying degrees through adjusting the coefficients; and introduce a remediation technique based neighbour area to reduce illumination influence and enhance image details, so the resource of the image which transformed has been improved. Experimental results show that this method is effective in compensating facial images of over-lighting and over-shading, and it is beneficial to improve the facial detection rate.关键词：illumination compensation;adaptive transformation;homomorphic filtering;remediation technique based neighbour area;logarithm transformation;face detection3174|206|0更新时间：2024-05-07
- 摘要：According to the characteristics of wheeled mobile robots, a novel method for real-time local two dimension(2D) map building based on binocular vision is proposed. Firstly, the principle of virtual height lines imaging in binocular vision system is proposed. The scene is partitioned into uniform grids, and virtual height line segments(VHLS) which do not exist in the scene are introduced. Each VHLS was projected into the stereo images in binocular vision system and formed a projection line segment(PLS).Finding the height of the scene is converted to finding the correspondence that has the maximum similarity metric on the PLS and in the searched horizontal disparity range. Then, a novel local 2D map building method is proposed. Similarity metric of the scene on the part of the PLS whose height is greater than the maximum which robots can overleap, is to determine whether the scene grid is an obstacle region or not. The experiment results show that the proposed method is valid and real-time for mobile robots, and can be used for building three dimension(3D) map.关键词：binocular vision;two dimension map;virtual height line;robot navigation2724|233|0更新时间：2024-05-07
- 摘要：During the traditional wavelet rotation-invariant texture retrieval algorithms, the extracted directional information is limited and the inter-scale dependency between the coefficients is ignored, which affects the efficiency of retrieval. In this paper, the authors propose a novel progressive rotation-invariant texture retrieval algorithm based on inter-scale dependency. Firstly, Log-polar transform and Non-subsample Contourlet transform (NSCT) are combined to acquire rotation-invariant multi-scale and multi-orientation coefficients, then generalized Gaussian distribution (GGD) model is used to extract the global structure information from low-pass coefficients which can be employed further as coarse retrieval features. Afterwards, the Non-Gaussian Bivariate Model is employed to model NSCT coefficients inter-scale dependency, which can be used as fine progressive retrieval foundations. Finally,the performance of the algorithm proposed is illustrated by experiments based on Brodatz standard texture database. Compared to inner-scale model GGD based on wavelet coefficients retrieval algorithm, our method provides better efficiency and accuracy, which is proved to be an efficient rotation-invariant texture retrieval means.关键词：inter-scale dependency;rotation-invariant texture retrieval;bivariate model;Log-polar transform;nonsubsample cntourlet transform3502|233|0更新时间：2024-05-07
- 摘要：Based on the geometry invariance of the rigid target during its 3D motion, the targets unknown 3D shape and motion can be reconstructed using scatterers of the 1D range extracted from the range image sequence of the single-antenna radar. For this geometric reconstruction of 1D-to-3D, we propose a 3D reconstruction algorithm for the shape and motion of the radar rigid target. The reconstruction can be realized using the complex 1D range data of the scatterers and the bundle adjustment for the target's reconstructed parameters is achieved with the nonlinear optimization utilized. Furthermore, since the targets translation model is introduced in the algorithm, the translation parameters can be solved together with the rotation. Thus, the impact for the reconstructed accuracy caused by the error of the range alignment processing is weakened. The simulations verify that, due to the remarkable improvement of the scatterers quantity and the richness of the target's 3D motion in the reconstructed data and, specially, the successful application of the optimization, the robustness of the algorithm is enhanced effectively.关键词：3D reconstruction;radar 3D imaging;geometry invariance;ISAR;bundle adjustment3010|191|0更新时间：2024-05-07
- 摘要：A new approach of image matching is introduced for 3D registration of augmented reality system. First, mathematical morphology is used to extract the edge of video images. And the land surface visible portions of DEM are derived at the viewpoint based on GIS visibility analysis. Second, the local extremes in the corresponding contours are detected and their curvature sets are calculated, respectively. Third, the similarity of the corresponding curvature sets is calculated by comparing the similarity matrix.关键词：augmented reality systems;video images;DEM raster images;image matching2698|240|0更新时间：2024-05-07
-
Helix rotation screen and the generation algorithm of helix slice in volumetric-swept display system
摘要：In order to meliorating defects, such as smaller image space and fewer voxels and severer voxel overlap dead zone caused by planar rotational screen in volumetric-swept display system, the system adopted helix rotation screen to construct an imaging space in volumetric-swept display system. Considering the characteristics of the imaging space, the generation algorithm for helix slices of the 3D model was proposed.It calculated the points of intersection crossed by simple-lines of the helicoids and triangular elements of 3D model and obtained the helix slices. This algorithm dispersed the 3D model into a set of consecutive helix slices according to the angle information of the rotational helix screen. The experimental results showed that the helix rotational screen performance was superior to plane rotational screen, and the proposed algorithm is convenient and effective.关键词：true 3D display;volumetric 3D display;volumetric-swept;helix rotational screen;helix slice2779|183|0更新时间：2024-05-07 - 摘要：Ratio of exponentially weighted average (ROEWA) operator is an multi-edge detector suitable for SAR images. The detector which is unable to compute the edge direction, cannot get the exact edge position. The reasons for these problems are analyzed and the flow chart of edge detection is modified in this paper. The constant false alarm rate (CFAR) character is retained by the traditional operator and the accuracy of non maxima suppression (NMS) is improved by the difference operator. Then Radon transform is carried out to get the edge direction for its excellent geometrical analysis and noise immunity properties, and by this way, the line connection is facilitated. The experiment results based on the SAR images show that the performance of the improved algorithm is robust and efficient.关键词：SAR;edge detection;ratio of exponentially weighted average;edge localization;edge direction3656|245|0更新时间：2024-05-07
- 摘要：A novel active contour model driven by statistical and gradient information is proposed in this paper. The model not only efficiently utilizes the gradient information of an object, which is in favor of fast and accurate location of boundaries, but also makes full use of the statistical information, including the global and local region information, which makes our method robust to noise. The use of the local region information makes the method free from intensity inhomogeneity of images, and the use of the global information helps to avoid the evolved contour trapping into local minima. Therefore, the initial contour can be set anywhere. Finally, the level set function is regularized by a Gaussian convolution kernel, which avoids an expensive computational re-initialization or regularization of the conventional models. Experimental results show that the proposed method can accurately and efficiently segment the homogenous images, as well as the inhomogenous images, with the initial contour set anywhere. Furthermore, the model is robust to noise.关键词：active contour model;statistical and gradient information;intensity inhomogeneity;Gaussian convolution;initial contour2876|179|0更新时间：2024-05-07
- 摘要：Rendering of under seawater scenes is of great significance for games, marine surveying, marine drive, and disaster succor. Its hard and time-consuming to real-time render the under seawater scene with caustics, shafts and scatterings because of the complex interaction between light and seawater. We propose consecutively an efficient method to render caustics based on the combining of wire-frame mode and gauss filter; a more realistic scattering phase function for under seawater to model the scattering and calculate the seawater color; a cylinder-based shaft algorithm to simulate the shaft light under seawater. Eventually, we introduced the hierarchy optimize method and GPU accelerate technology to successfully render different under seawater scenes in real time.关键词：under seawater scene;shaft light;scattering;caustic;rendering3714|246|0更新时间：2024-05-07
- 摘要：It is very challenging to construct a personalized virtual hand because of the complex shape and rich details of a human hand. An approach is proposed based on stereo vision and RBF interpolation to construct a personalized virtual hand model. Contour points and SIFT points are integrated to depict the features of a human hand, while a three-dimensional point cloud is extracted via stereo vision. A two-pass geometry alignment is performed to match a three-dimensional hand template to the point features. Guided by the point features, RBF interpolation is employed to deform the hand template to produce the personalized virtual hand model. Experimental results show that the proposed approach is functional enough to reconstruct the geometry of user’s hand.关键词：hand modeling;stereo vision;RBF interpolation;iterative closest points;personalization2981|220|0更新时间：2024-05-07
- 摘要：In this paper, we provide a new rhythm translation method based on Gaussian process dynamical model which can translate existing rhythm information to various types of motion accurately and effectively. Firstly, Short-Term PCA is used to calculate rhythm points in source rhythmical motion, three important motion features are combined to extract the motion feature points in target motion; Secondly, dynamical programming is utilized to find the best match between them in order to reduce the estimation time and damage to the target motion; Lastly, the Gaussian process dynamical model is applied to learn the target motion and rhythmical interpolation is done in latent space to get the new final rhythmical motion.关键词：machine learning;motion rhythm;motion synthesis;interpolation3252|187|0更新时间：2024-05-07
- 摘要：In the past several years, for displaying high resolution images on a limited screen, image retargeting has obtained more attention in the academic and industrial communities. Plenty of previous work used different importance maps to acquire optimal image resizing results. In order to realize the objective, which is to dispose the valueless information, of retargeting,we employ an image information redundancy analysis to guide the resizing process. First, we introduce an image decomposition framework to separate the high gradient redundant regions. Thereafter, we introduce a texture gradient to analyze the texture redundancy. Lastly, we show that the proposed texture gradient is also suitable not only for cartoon image analysis, but also for applying existing retargeting process to deal with non-photo realistic photos. Experimental results show that insensitively to the retargeting framework, our method is quite robust of the images with complex scenes or of multiple objects.关键词：image retargeting;salience map;texture analysis;cartoon image retargeting2723|268|0更新时间：2024-05-07
- 摘要：Decimation and compression of pre-computed deformation data is crucial to synthesize realistic animation of massive plants. In this paper, a motion simplification algorithm is proposed based on the observation that plants are of tree-like hierarchical structure inherently and move in a hierarchical mode. According to the hierarchy of a plant, the motions of its sub-trees in each depth are decomposed into low-frequency motion and high-frequency motion. The former denotes the dominant transformation of the plant within that sub-tree range and the latter contains residual deformation details. By breadth-first traversal of the hierarchy, the motion composed with low-frequency motions of all the sub-trees at a certain depth is the approximation of original motion at that depth. While traversal depth increased, the approximated motion would approach the original motion progressively and become more realistic, which results in a kind of levels of motion detail . The experimental results show that the presented scheme can simplify and compress the plants motions efficiently, and is convenient for the realistic animation of a large-scale forest.关键词：plant;computer animation;geometric simplification;levels of motion detail2794|189|0更新时间：2024-05-07
- 摘要：The existing methods of mesh fusion for 3D models are based on geometric mesh, which needs sufficient number of vertices to enhance the purpose, to connect components. However, the 3D characters in current games and animations generally use low poly model, which mainly uses a variety of maps to show details. To befit the situation of the low-model characters sparse vertices and extensive use of maps, we propose a simple method of harmonic mapping and averaged-plane to simplify mesh connection process and improve the efficiency. Thereafter, map synthesis based on affine transformation is applied to solve the problem of generation of the new map. Therefore, the new seamless fused model is acquired. Based on methods used in this paper and operational processes of model design and production, a Human-Computer interactive fusion application is developed to implement the efficient 3D model fusion and map synthesis operations. Many experimental results have shown that this approach is fast and effective enough to accomplish the fusion for 3D-character models in games and animations, forms a new character model that is more vivid, and thereby makes more interesting applications.关键词：3D model fusion;harmonic mapping;map synthesis;averaged plane2926|183|0更新时间：2024-05-07
- 摘要：Based on the Morse theory, a novel approach to create quad-mesh multiresolution from triangular mesh models is proposed. A smooth Morse function on a given triangular mesh is firstly defined as the solution of a Laplacian equation with constraints in which critical points are either specified by user or exacted from an eigenfunction of the Laplacian matrix of the mesh．As critical point layout has been carefully treated, maximum, minimum and saddle points of the function will automatically possess of structure of a quad mesh whose connectivity is then produced by tracing the stream lines of the function guided under the gradient field of the function. A critical point exchange rule is employed to generate the structure of the next finer quad mesh whose connectivity is finally also generated through stream line tracing. Parameterization is not required in the process as all resolution levels are created using the stream lines method．关键词：quad-remeshing;multiresolution;Morse theory;Laplacian field2668|196|0更新时间：2024-05-07
- 摘要：In this paper,a new unitary transform called U transform is constructed with the recursion method.The U transform contains piecewise constant basis vectors,piecewise linear basis vectors and piecewise polynomials of degree 2 basis vectors,and so it is a generalization of Walsh transform and Slant transformThe corresponding fast algorithm based on Kronecher product is deduced.The new orthogonal U transform is applied to digital image compression with quantization table based on Human Visual System(HVS).The experimental results indicate that the compression performance of U transform outperforms the Slant transform and is equivalent to that of DCT.关键词：image compression;orthogonal transform;Kronecher product;fast algorithms3202|199|0更新时间：2024-05-07
- 摘要：In order to improve the robustness of image Hash algorithm a new image Hash algorithm based on discrete curvelet transform is proposed. The image is firstly preprocessed, and then decomposed with discrete curvelet transform via wrapping. The curvelet coefficients of low frequency contained the main features of image and the coefficient of details of two layer contained rich edge information are as the feature vectors. And Logistic equation is used to encrypt the eigenvector. Finally, the image Hash sequence is obtained by quantization and compression. Experimental results show that the algorithm has better robustness compared to some other Hash method. It is fragility to different images. The chaos system enhances the security.关键词：image Hash;curvelet transform;image authentication2930|170|0更新时间：2024-05-07
-
Content-based remote sensing image retrieval based on feature selection and semi-supervised learning
摘要：Content-based remote sensing image retrieval has been receiving more attention. Current research on content-based remote sensing image retrieval is mainly about feature extraction and feature fusion. But all these methods have ignored such a fact that retrieval objectives of different types ought to adopt different features. Besides, small size of the training set is also a challenging issue. Considering the abovementional two issues, a new remote sensing image retrieval method, which is based on feature selection and semi-supervised learning, is proposed in this paper. The new method includes four steps: 1) Determine the number of clusters automatically by using MDL criterion; 2) According to a clustering validity index, we select the optimal features which can describe the retrieval objectives best, meanwhile make some improvements in the original DB index; 3) Dynamically determine the weights of the best color feature and the best texture feature; 4) Automatically select an appropriate semi-supervised learning method and conduct image retrieval. Experiment results show that, in contrast to the retrieval performance of relevance feedback method, the method proposed in this paper obtains similar retrieval performances both in the applications of soil erosion area retrieval and general land cover retrieval in remote sensing image, as well as solving the small sample size problem.关键词：content-based remote sensing image retrieval;semi-supervised learning;feature selection;clustering;cluster validity index3205|311|0更新时间：2024-05-07
- Postal code：100190
- Tel：010-58887035/58887030/58887418 Email：jig@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备05080539号-4
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0