最新刊期
卷 19 ， 期 9 ， 2014
- 摘要：Energy functional regularization model is an active research field in image restoration. To draw more attention from researchers of engineering community, and to push forward the research of the ill-posed problem, the paper provides a comprehensive analysis of recent development on regularization model of image restoration. Firstly, the relationship between total coordinate and local coordinate of image is established, and the principle of regularization term of image restoration is analyzed, and the isotropic and anisotropic diffusion theorems of regularization model are given. Secondly, based on functional space, image decomposition and tight frame, the disadvantages and advantages of the regularization models are analyzed, and the state of the art Methods are reviewed. Furthermore, the well-posed properties of solution of regularization model are analyzed. The basic diffusion principles of image restoration regularization model are deduced, the general formulas of regularization model are given, the potential problems and future development trends are discussed. Regularization technology plays a key role in opposite problem research, such as image restoration and inpainting, currently, though some excellent Results have been reported, many theory challenges still need to be investigated.关键词：image restoration;regularization model;partial differential equation (PDE);energy functional3127|304|0更新时间：2024-05-07
- 摘要：The development of steganalysis technology, especially universal blind steganalysis based on statistical characteristics, has increased the demand for steganographic security against statistical analysis. To ensure stego-image security against statistical analysis while avoiding overtraining to an incomplete cover model, this study presents a steganographic Method that minimizes embedding distortion. A novel distortion function reflecting higher order statistics is first defined based on element cliques. According to the Results of theoretical derivation and experiments, the maximal value of the Fisher criterion function is used as the optimization criterion for the parameters in the distortion function, such that the distortion function can be related to statistical detectability. Finally, when a secret message is embedded, multiple different feature subsets are integrated through Gibbs sampling and syndrome-trellis coding.Statistical characteristics are preserved while distortion is minimizes. Experiments are proposed to compare the classification errors of the new Method with those of three similar steganalysis Methods with different dimensions. Results show that the new Method can better preserve image model and maintains high security even when detected using a high-dimensional steganalysis Method. The classification error using corresponding feature set is higher than 0.4 while the embedding rate is 0.5 bit/pixel. The new steganographic Method successfully preserves statistical characteristics while minimizing distortion function. Moreover, the proposed steganographic Method effectively avoids overtraining to an incomplete model and has better adaptability and security than similar Methods.关键词：steganography;Gibbs random field;distortion function;feature analysis3320|259|0更新时间：2024-05-07
- 摘要：Most image interpolations only consider the quality degradation of low-resolution images by down sampling without accounting for quantization noise. We propose a novel compression image interpolation Method using adaptive symmetrical autoregressive models in this paper. Blocks with similar local images are assumed to have the same interpolation model.The proposed Method has two phases: training and reconstruction. In the training phase, the local gradient direction is first obtained via principal component analysis (PCA) to classify all training blocks into four directions and to build the corresponding symmetrical autoregressive models and training sets for every direction. Second, the training sets for every direction are classified into subclasses according to the basic features of the K-means clustering algorithm. Finally, the model of the direction to which each subclass belongs is chosen, and the constrained least square Method is used to estimate the weights of the model. In the reconstruction phase, the direction of the pixel is first determined according to the local gradient direction of the neighboring pixel. Subsequently, by computing for the Euclidean distance between local primitive features and every clustering center in the selected direction, the model of each subclass with the least distance is chosen for interpolation. The eight test images and two quantization parameters of JPEG still image compression are used in the tests. Results show that the proposed Method is better than other interpolations on the PSNR and SSIM, even in the presence of serious quantization noise. The Results demonstrate that the proposed Method produces better Results than other interpolations for both quantitative and visual comparisons and has low computational complexity.关键词：compression image;image interpolation;training;adaptive;symmetric autoregressive models4078|302|0更新时间：2024-05-07
- 摘要：The total variation image debluring model (ROF) based on energy functional opened up a new area of research on image processing, particularly on the application of partial differential equations. The defects of the ROF model have prompted many scholars to study the improved model and its algorithm. These scholars have achieved good Results. This study presents a TV image denoising model based on energy functionals and HVS. The existence of solutions for the denoising model in this paper is proven by using the comparison principle of the partial differential equation, and the Euler-Lagrange equation of the model is given using the variation principle. In terms of the numerical calculation of the model, this study discusses the discrete form of numerical approximation solution through artificial time algorithm, finite difference Methods, and numerous image denoising MATLAB experiments. Finally, the two indexes of noise quality are evaluated based on the peak signal-to-noise ratio (PSNR) and mean structural similarity (MSSIM). The (0.5~1) dB PSNR and the (0.05~0.3) MSSIM present an improvement based on the experimental data and Results. From the analysis of denoising, the TV image denoising model based on energy functionals and HVS can maintain the image edge and texture features and is thus superior to conventional TV denoising models.关键词：total variation;human visual system;image denoising;Euler-Lagrange equation;finite difference;peak signal-to-noise ratio (PSNR);mean structural similarity (MSSIM)4280|307|0更新时间：2024-05-07
- 摘要：Salt-and-pepper noise is one of the most common factors causing image contamination. Salt-and-pepper noise estimation has a guiding role in determining the size of the filtering window in denoising. Thus, we propose an algorithm based on the partitioning strategy to estimate salt-and-pepper noise density. The proposed algorithm horizontally and vertically splits the image equally into sub-blocks, counts the pixels of the sub-blocks with gray levels of 0 or 255, sorts all sub-blocks, selects candidate sub-blocks according to the characteristics of the different equences of the sorted pixel numbers, and uses the median of the noise density estimations of all the candidate sub-blocks as to estimate the noise density of the whole image. To evaluate the proposed approach, two different types of images are processed using the presented Method, and the noise density estimation Results are compared with those of existing salt-and-pepper noise density estimation algorithms. Simulation Results show that the new algorithm can accurately estimate noise density under different intensities and is effective for images that have many extreme pixels with gray levels of 0 or 255.关键词：salt-and-pepper noise;density estimation;median filter;histogram3279|332|0更新时间：2024-05-07
- 摘要：Feature extraction is the most critical step in pattern recognition, and facial expression recognition is no exception.Weber Local Descriptor (WLD) is a Method that can effectively extract texture information from images and has the advantages of being consistent with human perception of human beings and being insensitive to noise and non-monotonic illumination variations. However, WLD has some limitations in the feature representation of local details. To overcome these limitations, a facial expression recognition Method based on Pyramid WLD (PWLD) is proposed in this study. First, facial images are preprocessed. This step includes the detection of faces from facial expression databases and normalization. The salient regions of segment 2 that have significant contributions to facial expression recognition from images are also preprocessed. One of these salient regions is that which includes the eyes and eye brows, while another is that with the mouth. The sizes of salient regions differ, and these regions contain different information. Thus, we stratify these salient regions and divide each layer into different blocks. The PWLD features of each block in each layer are then extracted and cascaded to represent the global and local features of a salient region reasonably, with some parameter adjustments. Second, we compute for the Chi-square distance of the PWLD histograms in both the testing and training sets. We then choose the minimum distance in every category of expressionsand normalize this distance to construct the Basic Probability Assignment (BPA) as independent evidence. To create the BPA, we use curve fitting in numerical analysis by simulating several sets of data. Finally, fusion BPA is obtained by using the Dempster-Shafer rule, and the Results are further obtained by employing thedecision-making and judgment of Dempster-Shafertheory of evidence. By fusing the PWLD features of the two different salient regions with Dempster-Shafer theory of evidence, we can overcome the limitations of a single regional featureand acquire more reliable and accurate Results. We conduct some cross-validation experiments on the JAFFE and Cohn-Kanade facialexpression databases, and the average recognition rates reach up to 95.81% and 97.47%, respectively. In addition, we perform some experiments with other algorithms, such as LBP, LDP, and Gabor; we also conduct some comparative experiments that combine the PWLD with different classifiers, such as 1-NN and SVM. The WLD, which is known as a robust image descriptor, can well extract the texture information of images. Moreover, the PWLD can accurately describe the local details, which have more advantages than the WLD features. The comparative Results of some typical Methodsverify the effectiveness and fault tolerance of the proposed Method. The proposed Method has certain robustness under simultaneous noise and light conditions.关键词：facial expression recognition;Weber local descriptor (WLD);pyramid Weber local descriptor (PWLD);Dempster-Shafer(D-S) theory of evidence3955|422|0更新时间：2024-05-07
- 摘要：It's hardly to directly detect the foot plants from motion capture data. Many previous works have already successfully found the foot plant constraints. However, none of these Methods is completely automatic without any interaction. In this paper, we present a foot plant detection Method based on spectral clustering for motion capture data. First, samples are represented by motion features of performer's feet. Second, parameters are selected by analyzing the norms of samples. Finally, foot plants are detected by spectral clustering algorithm. After applying on a blending motion data set, high accuracy rates of foot plant detection are acquired，the accuracy rates of foot plant detection can reach 98.72%. The analysis of experimental Results and the comparison with baseline Methods demonstrate the generality and efficiency of our Method.关键词：computer animation;motion capture;motion editing;foot plant detection;spectral clustering3239|240|0更新时间：2024-05-07
- 摘要：Most trackers base on sparse representation-based trackers consider only the minimal reconstruction error of the holistic representation or local features without fully utilizing sparse coefficients or ignoring the discriminant of dictionaries. Thus, these trackers share a high possibility of failure when a similar object or occlusion is present in the scene. Thus, this study proposes a novel tracker based on sparse appearance model with discriminative dictionary and weighted features (SPAM-DDWF). First, the proposed algorithm introduces the Fisher discriminative dictionary. We then use the discriminative dictionary to distinguish the target from the background. The weighted alignment-pooling based similarity measurement is proposed to locate the target accurately and handle the occlusionfinely. Furthermore, we employ a reconstruction error-basedupdate strategy of the weights of the sparse coefficients. This strategy adapts to changes in the appearance of the targetand reduces the possibility of a drifting problem when occlusion occurs. Compared with several state-of-the-art trackers on most benchmark sequences, the proposed tracker maintains a higher success rate and lower drifting error in scenes with illumination changes, complex background, and occlusion.The proposed tracker reaches a 76.8% average success rate and 3.7% success in decreasing the drifting error. Results indicate that the proposed SPAM-DDWF tracking algorithm performs accurately, effectively, and robustly, especially when the object is occluded by analogues.关键词：discriminative dictionary;local features;sparse coefficient;sparse representation;similarity measurement3593|468|0更新时间：2024-05-07
- 摘要：Eaves are precious Chinese heritage which possess profound historical and cultural significance. However, eaves' characters images have obvious characteristics of high wear resistance, high noise and complex topology. In order to achieve eaves' characters recognition and make a contribution to digital protection of cultural heritage objects, a new Method based on improved gradient vector flow field is proposed to extract skeleton of eaves' characters. The algorithm proposed in this paper is based on classic level set Methods which are usually implemented by fast matching Method. What is different is that traditional medial function of level set Method is replaced by gradient vector flow based medial function, which is more automatic and accurate. It is also the innovation point of this paper. The new algorithm is mainly achieved by two wave propagations. Experiments verify the effectiveness and accuracy of the algorithm. The skeleton is 98.03% similar to the standard skeleton of specific models constructed by Matlab2012a. when Gaussian, multiplicative, salt and pepper noise are added into the image, we got a skeleton 99.15% similar to the former one without any noise in it. Improved level set Method surpasses Hilditch thinning algorithm and distance transform algorithm and gets the best skeleton of eaves' characters, which is homotopy, thinness, centered and smoothness. The Results of the experiments indicate that our algorithm proves to be a useful and effective technique to extract skeleton of 2D objects with complex topology.关键词：eaves tiles;heritage conservation;centerline extraction;level set method;gradient vector flow3333|272|0更新时间：2024-05-07
- 摘要：Traditional Euclidean distance based complex network is usually sensitive to the non-rigid transformation of the shape image. To overcome this problem, in this paper, a novel shape feature extraction Method based on complex network model and relative coherent distance is proposed. First, an initial complex network is constructed with nodes corresponding to the boundary points and edges allocated with relative coherent distance as weights existing between each node pairs. Then, this initial network is threshold evolved to generate a series of sub-networks. At last, some topological features are extracted from these sub-networks to generate the feature descriptors for the shape image. Promising experimental Results on classification and retrieval show that the proposed Method has strong capability in discriminating and recognizing variety of object shapes. Comparing with the traditional distances, the relative coherent distance is more robust to the shape non-rigid and elastic transformations.关键词：complex network;relative coherent distance;inner distance;dynamic evolution3547|280|0更新时间：2024-05-07
- 摘要：A new Method for line matching is presented in this paper. It utilizes binary relations of geometric attributes. Four procedures are introduced in sequence as follows. First of all, it's necessary to define a series of binary relations between two segments and calculate the local similarity of two line segment pairs respectively from two images. The second step is to filter preliminarily out the local similarity with threshold values and get the candidate segment sets by stepwise Cartesian product operation. Thirdly, the global similarities of all of candidate segment sets need to be calculated. Finally, the final matched line segment sets can be obtained based on the local and global similarities. The Method is insensitive to the changing of line segment's endpoint position and length and the selection of two threshold values. Also it's robust and invariant to rotation, scaling and translation. In the meanwhile, it takes line segment's directions into full consideration, thus improving the matching accuracy. Algorithmic analysis and experimental Results have proved that the Method is reasonable, efficient and reliable.关键词：line matching;binary relation;geometric attribute;cartesian product3074|245|0更新时间：2024-05-07
- 摘要：Three dimensinal face reconstruction from single image is a classical problem, and has found its use in video game production and film production. The recent demands have upgraded to reconstructing high-resolution 3D face with high computational efficiency and accuracy. However, monocular 3D reconstruction is an ill-posed problem in nature. Meanwhile, high-resolution 3D reconstruction involves great computational cost. Aiming at these problems, we propose a Method to achieve accurate and efficient high-resolution 3D face reconstruction from a single image. The proposed Method is mainly based on feature adaptation and Laplace deformation. A morphable model is built using a group of 3D face samples for feature adaptation. In detail, we select a set of salient 3D features from each 3D sample, then a new face can be obtained through a linear combination of these 3D feature sets. The parameters of the morphable model are the linear combination coefficients. In feature adaptation, a scaled orthogonal projection is used to model the projection from 3D space to 2D plane, and we adjust the projection parameters and the model parameters, so that the 2D projection of the face generated by the morphable model is as close as possible to the 2D facial features extracted from a given image. After model adaptation, the personalized 3D features are reconstructed by linearly combining 3D feature sets with the computed coefficients. To achieve high-resolution 3D reconstruction, we use the reconstructed 3D features to deform a generic 3D face of high-resolution. The deformation is based on Laplacian Method which uses Laplacian coordinates to represent the local geometric characteristics at each 3D point of face model. According to this Method, the Laplacian coordinates of all the points should remain unchanged after model deformation, and also, a set of 3D features should be very close to the reconstructed 3D features after model deformation. Then high-resolution and personalized 3D face can be obtained by solving a least square problem. At last, the realistic 3D face is obtained through texture synthesis. Content of main experiments: Experiments aim to test the accuracy and efficiency of the proposed Method by comparing it to existed Methods. To this end, we project a group of 3D facial feature sets onto 2D plane, and reconstruct 3D features from these 2D projections. The accuracy of feature adaptation is evaluated by comparing the reconstructed features with the true features. The execution time is recorded to represent the efficiency of feature adaptation. Then, we use 3D features extracted from a set of 3D faces to deform a generic 3D face, and compare the deformation Results with the original faces to evaluate the effect of model deformation Methods. At last, texture mapped realistic 3D face are demonstrated. Compared with alternating least square Method, the proposed feature adaptation Method converges faster with lower reconstruction error because our model fitting process is guided by the necessary condition of the optimal solution. Compared with interpolation Method based on radial basis function, the Laplacian Method can deform face model in more reasonable way so that the reconstructed faces are closer to the original faces. This is because Laplacian Method maintains local geometric characteristics at every 3D point during face deformation, thus leads to better Result than interpolating using Euclidean distance metric. Also, the textured 3D faces have decent visual appearance owing to the high resolution of the face. This paper proposes to combine feature adaptation and Laplacian deformation for high-resolution 3D face reconstruction. Quantitative experiments indicate that our Method has better efficiency and accuracy compared with existed Methods. Also, qualitative experiments demonstrate impressive visual effect of the textured 3D face. Therefore, the proposed Method is very suitable for high-resolution 3D face reconstruction.关键词：morphable model;feature adaptation;Laplace method;texture synthesis;3D reconstruction3894|334|0更新时间：2024-05-07
- 摘要：The traditional mean shift tracking algorithmoften loses the targetwhen the object is similar to the background and lacks an effective model update strategy. To address these challenges, this study proposes a new visual object tracking Method. First, coefficients based on the color histograms of the background pixels around theobject are computedand introduced to the target model to reduce the location error. Second, sub-modelsare selected and updatedaccording to the matchcontributing degree in the current frame. Results show that the proposed Method can well restrain background distractions, while effectively updating the target model and eliminating the driftingphenomenon. This paper proposes avisual object tracking Method based on weighted background and selective sub-model update strategy to improve the traditional mean shift tracking algorithm in terms of object modeling and model update strategy.Results show that the proposed Method is effective and robust.关键词：object tracking;mean shift;weighted background;selective sub-model update3278|269|0更新时间：2024-05-07
- 摘要：When a Bézier curve is used to describe complex shapes, the problem of joining curve segments smoothly has to be solved. To maintain the continuity of the whole curve, adjacent curve segments must meet strict continuity conditions. A higher the requirement for continuity usually causes conditions to become more complex and involves a larger number of control points. This study improves the modifiable Bézier curve in the literature to achieve smooth connection between curves automatically and to construct a piecewise composite curve with numerous merits. We first present a sufficient condition of continuity for two curves with continuous position.On the basis of the sufficient condition , we prove that the modifiable Bézier curve can achieve a smooth connection under conditions that usually guarantee continuity only for the usual Bézier curve and most Bézier-like curves in the literature. We then use a transition matrix to convert the modifiable Bézier basis to a new set of basis functions. We employ this set of basis functions to define a new kind of curve according to the definition mode of the standard Bézier curve. We then analyze the smooth connection conditions of the new curve.Considering theses smooth connection conditions and by using a special definition mode, we construct a kind of piecewise composite curve. The connection of the control points between adjacent curve segmentsis apparently similar to that of the classical B-spline curve. However, the connections are actually different. The B-spline curve only has one edge between the control polygons of two adjacent curve segments, that is, only one control point is different. Nevertheless, the composite curve defined in this study only has one edge. Thus, only two control points are the same. The new curve defined by the new basis function has relatively simple and special smooth connection conditions. Two neighboring curves can be smoothly joined automatically as long as the last control edge of the former coincides with the first control edge of the latter. Furthermore, the degree of smoothness at the meeting point can be freely adjusted by simply changing the value of the parameter. The piecewise composite curve of the new curve possesses numerous desirable properties, such as geometric invariance, symmetry, automatic smoothness property, and local control capability similar to that of the classical B-spline curve. However, the main difference between the newly defined piecewise curve and the standard B-spline curve is that in the new composite curve, each segment can be defined by different numbers of control points. By contrast, the number of segments must be equal forthe usual B-spline curve.This difference is the main reason that the new piecewise curve has a stronger local control capability than the B-spline curve. Furthermore, a suitable parameter can enable the new composite curve to achieve the expected smoothness at each connection point. In addition, changing one control point can alter the shape of two curve segments. When the parameter of one curve segment is changed, only the shape of the current curve and the degree of smoothness at two connection points will change. This study presents the general Method to construct curves. This Method can easily achieve a smooth connection. We further present the general Method to construct composite curves. This Method can achieve a smooth connection automatically.关键词：curve design;Bezier curve;smooth stitching;piecewise combination;continuity3768|300|0更新时间：2024-05-07
- 摘要：The segmentation of pulmonary airway tree serves an important function in the accurate localization and quantitative evaluation of lung diseases. Considering the size, shape, and density variations among different bronchus branches, we propose a Method for 3D pulmonary airway tree segmentation by combining region growing and morphological grayscale reconstruction algorithms with a focus on solvinglocal leakage problems. First, the region of interestor maskis defined by extracting the lung parenchyma with a global threshold and morphological closingoperation.The trachea and large bronchi are then segmented by using an improved region growing Method based on the iterative hysteresis threshold, and a local volume explosion index is adopted to suppress the lateral leakage. The smaller bronchi are segmented using a 3D morphological grayscale reconstruction algorithm, and a connected component-based shape descriptor is defined to remove the pseudo-tracheal regions. Finally, a complete airway tree is obtained by integrating the segmentation Results from the initial two steps. The algorithms are tested by using the publicly available data of the EXACT'09 challenge, and quantitative evaluation is conducted with the bifurcation number, branch number, and branch ratio o fthe 20 test CT cases by comparing the details with a manual reference. Results show that the proposed Method can identify more than half of the branches in the reference, and the mean number of detected branches can reach 55.5% under a relatively low leakage rate. Compared with that of other considered Methods, the performance indexes of the proposed Method are moderately better. Although the algorithm appears to be simple and has low computational complexity, the proposed Method effectively prevents leakage.关键词：airway tree segmentation;region growing;morphological reconstruction;CT image3948|303|0更新时间：2024-05-07
- 摘要：Focusing on the traditional template matching Method for nodule detection, a 3D adaptive template matching algorithm for lung nodules detection has been proposed in this paper. First of all, 3D lung parenchyma is segmented from the scanned CT images, and then, canny operator is employed for extracting 3D ROI which will be used as the candidate pulmonary nodule. Secondly, collect the main direction of the 3D ROI and locate the center slice, and expand the 3D adaptive template from the center slice trace along the main direction. At last, calculate the correlation coefficient between the 3D adaptive template and pulmonary nodule candidate image by applying Normal Cross Correlation (NCC) algorithm, and set a threshold value of NCC for marking the higher correlation coefficient regions as detection Results. Based on 66 clinical cases' experiment, the sensitivity reaches 95.29% and false positive is 12.90%. The experiment Results show that our Method has high sensitivity and accuracy and can help radiologists to detect nodules effectively.关键词：lung nodule;computer-aided detection (CAD);template matching;3D adaptive template3976|281|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