最新刊期
卷 16 ， 期 11 ， 2011
- 摘要：The depth-of-field effect can improve the realistic sense of a computer-generated image,and thus rendering of this effect has become one of the important research topics in computer graphics.Aiming at the problem for rendering the depth-of-field effect,in this paper we first study optical imaging principles which various kinds of camera lens models featuve,and compare their characteristics and capabilities of modeling the depth-of-field effect;then the taxonomy of rendering techniques for the depth-of-filed effect is presented.Relevant fundamental principles and typical algorithms are concluded,and their advantages and disadvantages are compared in detail.Lastly applications of rendering techniques for the depth-of-filed effect are presented and future research trends are foreseen.关键词：depth-of-field;camera lens model;depth-of-field rendering approach;realistic rendering;real-time rendering3331|266|0更新时间：2024-05-07
- 摘要：The index map after vector quantization has a strong statistical correlation.That means the neighboring indices are the same or the offset between them is very small.Codebook sorting can,according to some criteria,enhance the correlation among neighboring indices.Based on the squared Euclidean distance between code words,a new codebook sorting method is proposed.Compared with the conventional mean-ordered codebook,the distance-ordered codebook has a much higher correlations between neighboring indices and the offset become even smaller.As a result,distance-ordered codebook can also significantly improve the compression efficiency of the AICS (adaptive index coding scheme) algorithm.关键词：vector quantization;index map;distance;codebook sorting2905|183|0更新时间：2024-05-07
- 摘要：In order to effectively preserve edges of low signal-to-noise ratio images,a kernel method-based selective anisotropic diffusion denoising algorithm is proposed.The algorithm is based on the anisotropic diffusion model of the multiphase hierarchy segmentation method.Because the image data is generally non-linearly separable,the data term of the multiphase hierarchy segmentation method is promoted from low-dimensional space to high-dimensional space by the kernel method.In the high-dimensional space the multiphase hierarchy segmentation method is applied for the image segmentation.Then,the diffusion coefficient of the P-M model is improved based on gradient information of the homogeneity region.Finally,the proposed P-M model is used to smooth noise in the homogeneity region.The experimental results show that the proposed algorithm can efficiently reduce noise while preserving edges.关键词：image processing;image smoothing;anisotropic diffusion;kernel method2869|231|0更新时间：2024-05-07
- 摘要：In this paper we introduce a new function for image denoising. The new function is simple and continuous. It obtains some advantages from both: the hard-thresholding function and the soft-thresholding function. It overcomes the shortcoming that there is an invariable dispersion between the estimated wavelet coefficients and decomposed wavelet coefficients of the soft-thresholding method. At the same time, this function overcomes the shortcoming of the hard-thresholding method with discontinuous functions. We proof that the new function satisfies the shrinkage condition and has infinite rank continuous derivative. At the same time, it has adaptive character and is suitable for various mathematical processing. These advantages make it possible to construct an adaptive algorithm for image denoising. At last, several numerical experiments show that the proposed new function is very effective and predominant. It gives better performance both in terms of PSNR and in visual quality. Our function can preserve more significant information of original images like edges and details than the soft-thresholding function. At the same time, the images denoised by our function are smoother than those denoised by the hard-thresholding function. It also gives a better MSE performance than two typical methods.关键词：wavelet transform;hard-thresholding function;soft-thresholding function;new denoising function;MSE estimate;image denoising2918|262|0更新时间：2024-05-07
- 摘要：Based on the polarimetric orientation angle estimation,we discuss the decomposition error caused by the azimuth slope. We analyze the azimuth slope effects on the Yamaguchi model,and use the polarimetric SAR azimuth estimation to reduce the terrain effect. Finally,we apply this approach to a full polarimetric AIRSAR image of San Francisco from 1992. The experiments show that the polarimetric orientation angle composition could improve the precision of the Yamaguchi model.关键词：polarimetric SAR;polarimetric orientation angle;azimuth slope;Yamaguchi model decomposition2719|212|0更新时间：2024-05-07
- 摘要：To achieve accurate 3D positioning and real-time tracking of moving objects,a video positioning method using multiple orthogonal cameras together with an iterative algorithm that approaches the coordinates of the 3D objects alternately is proposed.The optical axes of the cameras are set along axes of an orthogonal coordinate and all cameras are pointed to the origin of coordinate.Unlike the most of current computer vision method,the iterative algorithm does not involve image alignment operation,which could affect the positioning efficiency and precision.The convergence of the iterative algorithm is proved.Numerical examples and practical tests show that the proposed method is simple to compute,insensitive to errors,and has a rapid rate of convergence,thus has a great potential in application.关键词：computer vision;tracking;iterative methods;video signal processing;multiple cameras3119|154|0更新时间：2024-05-07
- 摘要：With the development of image processing technology, the image segmentation technology is also in maturity, however, more segmentation method has some limitations at present, it is very difficult to realize global segmentation with the traditional method, and difficult to realize the efficient and accurate segmentation for the objects with the weak or blurred edge. A novel region-based active contour model is proposed in this paper. It is based on the geodesic active contour GAC and C_V model. Through the experimental analysis, first, the algorithm greatly improves the accuracy of segmentation, the evolution of contour can stop near the edge of object, even if the edge of target is weak or blurred. This algorithm overcomes disadvantages of traditional active contour segmentation algorithm in local segmentation, effectively realized the global segmentation.关键词：geodesic active contour model;C_V model;level set;image segmentation3133|315|0更新时间：2024-05-07
- 摘要：Aiming at the problem of preserving the discontinuities in optical flow field for motion segmentation, a meaningful research has been done. The key of the research focuses on the improvement of the diffusion tensor in reaction-diffusion model. By analyzing the existing methods, we proposed a new method which design diffusion tensor by joint image- & flow- driven. Furthermore an efficient coarse-to-fine numerical scheme about the partial differential equation has been present at the same time. our experiments proved that our proposed approach can effectively overcome the motion blur problem of other methods while and preserving the discontinuities in the optic flow field more accurately.关键词：optical flow;motion segmentation;diffusion tensor;partial differential equations3111|217|0更新时间：2024-05-07
- 摘要：Due to the correction of the bias field,it is hard to obtain the accurate segmentation results of magnetic resonance(MR) images using traditional methods.In this paper,a set of basis functions is constructed firstly to fit the smoothness bias field;then the information of the bias field is introduced to the Gaussian density function,and according to the statistics classification rule,we define the energy function for the brain MR image segmentation and bias field correction.At last,this energy function is incorporated into a three-phase level set framework to propose our model.Compared with other approaches,our experiments demonstrate that our method not only can obtain accurate segmentation results but also can restore images better.关键词：bias field;three-phase;level set;image segmentation3535|232|0更新时间：2024-05-07
- 摘要：The amount of remotely sensed data increases rapidly,and the information contained in this data becomes more and more complicated,the way how to classify these datasets generalized and effectively is a problem which needs urgently to be solved.A modified rotation forest algorithm is proposed which takes the RBFNN as the base classifier to classify the remote sensing image.The input training dataset is changed by the rotation forest which can output a much small sub-feature.Then the non-redundancy feature set is got by using PCA technology to process these new sub-features.Finally,the training dataset changes according to the coefficient by the PCA transformation.This change will lead a higher diversity factor among these sub-classifiers which will give a much higher accuracy.The proposed method can obtain higher classification accuracy than other traditional methods when it used on the Zhalong wetland remote sensing image,and this algorithm has much higher generalization ability and much less over study phenomenon.关键词：hybrid algorithm;RBFNN;rotation forest;feature transformation2545|201|0更新时间：2024-05-07
- 摘要：A dimensional reduction method based on the factor analysis model is proposed for hyperspectral data to resolve the problems of high relativity of bands and large volumes of data.The intrinsic dimensions of hyperspectral data can be obtained by our method through further processing,including solving the factor payload matrix, computation of model parameters and rotated matrix,and the estimation of the factor contribution.Less composite factors can be found to replace data of all bands,which can not only represent almost information of original data,but is also factor independent.Push Hyperspectral Imager (PHI) data is used to evaluate the performance of our proposed method.The result illuminates Kappa parameter is improved from 0.744 to 0.821,and all useful information of data is reserved,relativity among bands is removed,and class separability is increased after dimensional reduction.关键词：hyperspectral remote sensing;dimensional reduction;factor analysis model;push hyperspectral imager(PHI)2846|197|0更新时间：2024-05-07
- 摘要：Aimed at the problem that nearest neighbor method and k-nearest neighbor method can't obtain better classification effectiveness when there aren't enough labeled examples,a semi-supervised classification method is proposed in this paper.The method is based on the mechanism that unlabeled samples were used if human classify pattern involuntary.The method utilizes the nearest neighbor relationship between unlabeled samples to reduce the influence of the number of labeled samples on classification accuracy.The experimental results using the MNIST database of handwritten digits and the ORL face database show the method has higher classification accuracy than the nearest neighbor method and the k-nearest neighbor method if there aren't enough labeled samples.关键词：nearest neighbor classification method;semi-supervised learning mechanism;semi-supervised classification3132|214|0更新时间：2024-05-07
- 摘要：Since Wireless Capsule Endoscopy (WCE) is a novel technology for recording the videos of the digestive tract of a patient,the problem of segmenting the WCE videos of the digestive tract into sub-images corresponding to the mouth,stomach,small intestine and large intestine regions is not well addressed in the literature.A few papers addressing this problem use a supervised learning approach that presumes availability of a large database of correctly labeled training samples.Considering the difficulties in procuring sizable WCE training data sets needed for achieving high classification accuracy,we introduce an unsupervised learning approach that employs Scale invariant feature transform (SIFT) with color information for extraction of local features and uses probabilistic latent semantic analysis (pLSA) model for data semantic analysis.Our results indicate that this method compares well in classification accuracy with the state-of-the-art supervised classification approach to WCE image classification.关键词：video classification;unsupervised learning approach;probabilistic latent semantic analysis (pLSA);Scale invariant feature transform (SIFT)2950|206|0更新时间：2024-05-07
- 摘要：Axis extraction of three-dimensional blood vessel images is the first and most important step in quantifying blood vessel.A dynamic method of extracting axis of three-dimensional blood vessel images was proposed.Firstly,the energy constraint equations has been constructed,the initial skeleton curve of blood vessel images obtained by thinning algorithm or artificial constructing method keeps the approximation location of the axis of blood vessels under energy constraint equations and along the distance field gradient direction of blood vessel images.When the equation energy reaches a minimum value,the initial skeleton is also fixed in the axis position at this time.Experimental results show that the position of the blood vessels axis extracted by this method is accurate,and the axis preserves topology and connectivity.关键词：3D blood vessel image;axis;constraint equation;thinning;distance transformation2484|195|0更新时间：2024-05-07
- 摘要：Regarding the characteristics of the anisotropic diffusion model, an efficient multi focus image fusion method is proposed using a rule of difference coefficients between anisotropic diffusion model and Gaussian filter. Anisotropic diffusion equation is used to filter an image depending on local properties of the image.The image is smoothed in the homogenous areas while image features are preserved effectively on edges. The resulting fused image is composed of adaptive pixels which are chosen directly from the corresponding original images according to a selection rule of high pass coefficients. Those high pass coefficients are provided by accumulated values over a square sliding window using difference images between anisotropic diffusion model and Gaussian filter. Experimental results demonstrate that the proposed fusion algorithm is very suitable for image fusion of multi focus images.关键词：image fusion;multifocus image;anisotropic diffusion;Gaussian function3163|211|0更新时间：2024-05-07
- 摘要：In this paper,the importance of cost aggregation,also called similarity measure,for belief propagation and the interaction of them are discussed.A global stereo matching algorithm is proposed by combining the belief propagation and local edge construction-based cost aggregation.First,a virtual closed edge is formed surrounding each pixel via second derivative operator in order to construct an adaptive window for the centered pixel.Then,the local cost aggregation is calculated on support pixels in an adaptive window.Finally,accelerated belief propagation optimization algorithm is used to obtain the disparity.The experiments based on the Middlebury benchmark indicate that the local edge construction-based cost aggregation can do well with belief propagation optimization and show encouraging results of the proposed stereo matching algorithm.关键词：stereo matching;belief propagation;cost aggregation;local edge construction3391|231|0更新时间：2024-05-07
- 摘要：This paper aims at symbol sharing between CAD and TrueType. First, point symbol model between CAD and TrueType were compared from data structure and storage style. Second, conversion methods between them were proposed and some issues such as data size, information distortion and lost on fore-and-aft conversion were analyzed. Experiment validates the feasibility of conversion between CAD and TrueType point symbols and it help to symbol sharing.关键词：CAD symbol;TrueType;symbol model;symbol conversion;symbol sharing2867|198|0更新时间：2024-05-07
- 摘要：This paper presents a method which is based on edge tracking algorithm to generate line drawings from images.The algorithm consists of two parts:edge tracking and line drawing painting.For edge tracking,we propose an edge tracking algorithm based on dissimilarity measure,so that the edges that be from the edge detection operator are able to be classified and connected.In the line drawing painting process,we employ a non-uniform B-spline to interpolate for the discontinuous edge and the Gaussian function to obtain a continuous smooth lines.Then the brush is generated based on the curvature of the lines,and the line drawing of images is obtined.The experiment results of image line drawings,which is generated by our method,are demonstrated in this paper,and experiment results show that our method is able to quickly generate higher-quality line drawings.关键词：line drawings;edge tracking;non-uniform B-spline;Gaussian smoothing;dissimilarity3636|250|0更新时间：2024-05-07
- 摘要：Traditional remote sensing image change detection approaches based on structure features are usually limited by imaging stability. In this paper, we introduce a new method for unsupervised change detection in remote sensing images using compressive sensing (CS) based on the image inherent sparse structures. For this algorithm, a large collection of image patches is projected onto high dimensional spaces through redundant dictionary, giving an adaptive sparse representation per each image patch. A random matrix is taken as measurement matrix to realize feature space dimension reduction. Then, the final change detection is realized by clustering the feature vector space using the fuzzy C-mean clustering(FCM)algorithm, achieving the reconstruction of change regional information. The experimental results demonstrate that the proposed algorithm has good change detection results both in contour and region and has a good robustness.关键词：change detection;remote sensing image;compressive sensing (CS);fuzzy C-means (FCM) clustering3242|422|0更新时间：2024-05-07
- 摘要：Gaussian mixture model (GMM) clustering algorithm is widely used in image segmentation during recent years. The algorithm is however quite sensitive to speckle noise since spatial correlations between pixels are ignored. This paper presents a region-based GMM clustering algorithm for SAR image segmentation featured by incorporating spatial correlations. The watershed algorithm is first used to generate primitive homogeneous regions. Regional mean values are then calculated as input samples of the GMM clustering process. The impact of noise on the segmentation result can therefore be reduced in the space of regions instead of pixels. A feedback mechanism is further introduced into the expectation-maximization (EM) algorithm to improve the precision of parameter estimation. The efficiency of the proposed algorithm has been demonstrated on the segmentation of synthetic SAR images and real SAR images, where the segmentation accuracy has been substantially improved in contrast to pixel-based the GMM algorithm.关键词：image segmentation;watershed;Gaussian mixture model;EM algorithm4618|248|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