最新刊期
卷 15 ， 期 12 ， 2010
- 摘要：In order to achieve better rate distortion (RD) performance in distributed video coding (DVC), a new rateadaptive low density parity check (RALDPC) coding algorithm is proposed. Firstly, a high compression ratio code is constructed based on progressive edge-growth (PEG) algorithm. Then low compression ratio codes are obtained using row splitting algorithm. The rate can be well adjusted by merging and splitting the parity-check matrix rows. And three restriction conditions for merging check nodes are proposed in order to obtain good performance codes. Based on the RALDPC, a rate adaptive DVC algorithm is proposed, which can achieve better RD performance about 0.1~0.7dB compared to DVC algorithm based on the popular Turbo codes, when the correlation between the original and side information is medium or high.关键词：distributed video coding(DVC);low density parity check (LDPC);rate-adaptive;progressive edge-growth(PEG) algorithm;syndrome3316|244|0更新时间：2024-05-07
- 摘要：Image super-resolution(SR) reconstruction refers to a signal processing approach which produces high-resolution images from observed multiple low-resolution images A new method for simultaneous image super-resolution and motion estimation is proposed to expand the application range of SR technology. The framework of SR resolution and motion estimation is given based on maximum a posteriori (MAP). The framework takes into account both the influence of HR image dispersion between two iterations, and the weight of different LR images, which makes the algorithm self-adapting. The framework then can turn to SR resolution and motion estimation model. Nonlinear least squares method is employed to solve the model to get the global motion area of SR resolution. Our experimental results show the effectiveness of the proposed algorithm.关键词：super resolution;maximum a posteriori(MAP);image enhancement;nonlinear least squares methods3103|425|0更新时间：2024-05-07
- 摘要：Image super-resolution reconstruction has been a hot research topic in recent years. Among kinds of reconstruction methods, regularized reconstruction is widely used, because it applies simple principle and unique solution. The regularization parameter plays an important role in reconstruction. If the parameter is too small, the noise will not be effectively restrained, conversely, the reconstruction result will become blurry. Therefore, a U-curve based reconstruction method is proposed, using the unique features of U-curve to select the regularization parameter. The data fidelity term and a prior item are used to form a U-curve function, and the left maximum curvature point is selected as the optimal regularization parameter. The proposed method is tested on two simulate data sets. The results show advantages of this revised method both in visual effects and quantitative evaluation.关键词：super resolution reconstruction;regularization;L-curve;U-curve3413|251|0更新时间：2024-05-07
- 摘要：According to the vision character of cone and rod in human’s retina, a new model called vision adaptability model is proposed in this paper, the model aims at enhancing color image. Based on Retinex and vision adaptability model a new algorithm for image enhancement is proposed, in this algorithm the illumination image’s approximate is simply removed and the reflectance image’s approximate is calculated, then the global contrast and brightness of reflectance image’s approximate are adjusted according to the proposed model, which make the enhanced image agree with human’s vision. The proposed algorithm and some classic Retinex algorithms are applied to the same degraded RGB images in the experiment, then qualitative analysis and quantitative analysis are conducted on the experimental results. It can be concluded from the comparison that the proposed algorithm performs better than preexisted Retinex algorithms on enhancing the image details and upgrading the global contrast.关键词：vision adaptability;Retinex;image enhancement3435|350|0更新时间：2024-05-07
- 摘要：When the rocket or missile launches, the ground testing images are interfered seriously by smoke. The research on the weakening of smoke in images is very important in the ground testing image preprocessing. This paper firstly discusses smog removing in the frequency domain, and it proposes a new homomorphic filtering method. Then it proposes Weighing Wavelet Coefficient approach weakening smoke on account of the multi-resolution and space-frequency nature of the Wavelet Transform. Finally it adopts five different objective appraisal methods to evaluate the disposed images. Experimental results show that the two proposed approaches are more effective on smog weakening than the other three.关键词：smoke weakening;smoke image;homomorphic filtering;wavelet transform3292|262|0更新时间：2024-05-07
- 摘要：Digital images have an inherent amount of noise typically uniform across the entire image is introduced by imags processing. So if images with different noise levels are spliced together would leave an evidence of tampering. Base on this characteristic, this paper presents a passive image forgery detection algorithm which uses blind estimation of background noise. By computing high order statistic characteristics of the background noise and estimating the neighboring overlap blocks, the algorithm could locate the forgery parts. Experiment results show its effectiveness in detecting forgery part in spliced images.关键词：image forgery;background noise;blind estimation2994|221|0更新时间：2024-05-07
- 摘要：The relationship between amounts of coefficients and order in radix is discovered through research of DCT principle and proven in combination with the nature of cosine function. On this basis, a generic generating algorithm for N-order (N=2, k>0, sic passim) integer DCT transform radix is presented, which we do not need to analyze floating radix corresponding to integer’s. Through rearrange variations of coefficient, the mid-polynomials are extremely regularity. The group of polynomials in arbitrary N-variable is resolved by designing a N-digits with M as radix implementing N-loops to exhaust all possible solutions. The experimental results show that the algorithm can find all available radix for arbitrary N×N integer DCT as long as the computing capacity is enough.关键词：DCT transform;integer;N-order radix;amounts of coefficients;order3565|176|0更新时间：2024-05-07
- 摘要：Traditional linear feature detection methods based on structureless algorithms of Beamlet transform are mostly used to detect simple line segments and curves, while fail to detect complicated edges in natural images. Wavelet transform has great advantages in point feature detection, meaning that it is good at detecting edge and details. In this paper we improve traditional methods with the help of wavelet. Meanwhile, energy function in traditional algorithm is improved and a new drawing linear feature rule is proposed in order to represent a dyadic square with at most one optimal Beamlet. First, image is decomposed into low frequency and high frequencies with wavelet to highlight edge detail feature; second, the edge image’s transform coefficients are obtained by Beamlet transform. Finally the coefficients are dealt with using the improved energy function and linear features are extracted following the new drawing rule. Experimental results show that without costing obvious extra computing time, our proposed method can extract complete and clear linear features in natural images.关键词：wavelet transform;Beamlet transform;linear feature4012|264|0更新时间：2024-05-07
- 摘要：When line edges are linked with gradient phase and location information, linking result is easily affected by noise, which produces broken line segments. To improve the robustness of line extraction to noise, adaptive phase partitioning principle and location constraint principle for edge linking are proposed through analyzing the statistical characteristic of gradient phase and location of line edge. Based on above principles, a novel line extraction method is proposed. Firstly, the Canny method is used to detect edges. Secondly, short lines are extracted and used to estimate parameters for line linking. Finally, line edges are linked using “phase principle” and “position principle” alternatively. Experimental results show that the proposed method extracts lines completely and precisely.关键词：line extraction;adaptive phase partitioning;statistical inference theory;edge location constraint2648|224|0更新时间：2024-05-07
- 摘要：Registration of multi-resolution,multi-source images is a difficult task.This paper proposes a novel registration algorithm based on SAM(square root arithmetic mean divergence)information of multi-scale Harris corners(CSAM for short).In this algorithm,the estimated transform parameters are obtained by extracting a multi-scale contour and detecting multi-scale Harris corner.Then CSAM is used as similarity measure function,and several optimized match points are obtained.The final registration parameters can be acquired by using least squares method. Registration of medical images with noise and multi-resolutions can be realized by this algorithm,calculation time is reduced and local extremum can be avoided due to reduce matching corner points which need not optimal search. Finally, experimental results show that this algorithm has the advantages such as high precision, high speed and good robust.关键词：image registration;multi-scale;Harris corner;square root arithmetic mean divergence(SAM)4055|361|0更新时间：2024-05-07
- 摘要：To reduce labors such as user input, this paper proposes a simple stroke-based iterative image matting approach, which only needs a few user scribbles to mark foreground and background pixels. It builds robust color models from multiple feature spaces of pixels, and further introduces sampling constraint functions to collect valid foreground and background samples. By combing samples information and local smoothness constraint, the method defines a global energy function on all unknown pixels. The optimal matte is accomplished with iterative energy minimization. The experimental results demonstrated that the method can produce accurate and visually smooth results and is efficient to handle texture-rich images or images with ambiguous foreground and background colors.关键词：image matting;discriminative Gaussian mixtures;alpha matte3211|231|0更新时间：2024-05-07
- 摘要：In this paper, a discriminative locally linear embedding algorithm on image recognition, which considers spatial relationship of pixels and class information in order to improve the performance of locally linear embedding (LLE), is presented. First, neighbor matrix, which is used to compute weight matrix, is constructed by adaptive image Euclidean distance, and features are reconstructed using the weight matrix. And then intrinsic lower-dimensional space of data is reconstructed. Finally, linear discriminant analysis is utilized to introduce class information to solve the defects that LLE can’t reconstruct test samples and classify. Experiments are carried on FRAV2D and ORL databases. Comparing our proposed algorithm with popular algorithms in face recognition, these results show that a discriminative LLE can keep the best manifold structure and class information, and improve the accuracy of face recognition.关键词：adaptive image euclidean distance;locally linear embedding;linear discriminant analysis;face recognition3352|189|0更新时间：2024-05-07
- 摘要：As cloud can significantly affect image qualities in the earth studies,cloud detection plays an important role in analyzing satellite images.Much work has been done on it all over the world,these methods are suitable for large area,but sometimes the precision can’t meet the requirement.Based on the theory of remote sensing and digital filtering，We conduct statistical analysis on a local area (21.0°N~27.0°N,117.0°E~123.0°E).Through the statistic,characteristics and thresholds that are useful for the target area are found,which are Ref1,Ref6 and BT31.As a supplement,it selects Ref and BT31-BT32>Th31-32 to detect cirrus and thin cloud. At the end,the result is analyzed by comparing it with the outcome of MODIS operation arithmetic,which found this method improves the precision of cloud detection.关键词：MODIS;cloud detection;digital filtering;statistical characteristic;the southeast China2836|179|0更新时间：2024-05-07
- 摘要：A statistical background subtraction technique is proposed based on clustering of temporal color/intensity. An un-supervised clustering method is proposed to model a background with serial of clusters. The unimodal or multimodal distributions of background are detected adaptively. We use a Gaussians model to simulate each cluster which prevents the estimation the parameter of mix of Gaussians model. The foreground will be detected by comparing the background possibility with a threshold. Experimental results show our approach has equal or better segmentation performance and is proved capable of real-time processing.关键词：unsupervised clustering;Gaussians model;adaptive3099|247|0更新时间：2024-05-07
- 摘要：Differ from traditional 3D reconstruction via stereo vision, this paper studies the reconstruction of straight horizontal lines in 3D space from single 2D omni-directional images. It demonstrates that, for symmetric non-central catadioptric systems, the equation of a 3D horizontal line can be estimated using only two points extracted from its image . By exploiting the peculiar property of horizontal line image in catadioptric system, line detection from single omni-directional image can be simplified. Meanwhile, a horizontal line reconstruction algorithm based on main-point image and non-main-point image is developed. Detailed experiments justify that, under the same precision of image point extraction, compared to present primary approaches of reconstruction from four image points, our methods is more straightforward and accurate.关键词：single image;catadioptric omni-directional image;horizontal line;3D reconstruction2902|257|0更新时间：2024-05-07
-
Human pose estimation algorithm based on segmentations of human head and shoulder in video sequences
摘要：To describe and estimate a human pose in video sequences, most methods use 3D reconstruction. These methods often need multiple cameras, have high computation complexity and need many limit conditions. Because of these limitations, this paper proposes a human pose estimation algorithm based on segmentations of human head and shoulder. In this algorithm, the positions of human head and shoulder are located, then the head pose is obtained from the characteristics of plane imaging of human head and the body pose is obtained from the characteristics of the silhouette of human shoulder. Moving human pose is estimated by both head and body poses. The experiment results demonstrate the validity and superiority of the proposed algorithm.关键词：estimation of human pose;motion recognition;computer vision;video sequences3446|254|0更新时间：2024-05-07 - 摘要：Reconstructing 3D depth information from 2D defocus images is one of the top important research topics in computer vision. However, existing methods need to change the camera parameters, such as the focal length of the lens, the distance of the focused image from the lens plane and the radius of the lens, to attain the defocus images of different blurring degree. Unfortunately, in some cases with high level of magnification cameras, any change of any parameter will destroy the cameras drastically, so the application field of many existent algorithms is strictly restricted. Therefore, in this paper, a novel Depth from Defocus (DFD) method is proposed to solve this problem. First, two different blurred images are captured through changing depth. Second, the relation between depth and blurring is discussed based on the blurred imaging model obtained from the concept of relative blurring and the diffusion equation. Finally, the depth reconstruction is completed by solving an optimization problem. This proposed algorithm which does not need change any camera parameters or compute the focus image is easy to be realized. What’s more, the results of simulations and error analysis show that this method can reconstruct depth information with high precision and can be used in micro/nano manipulation and fast detection which are sensitive to camera parameters.关键词：depth from defocus;diffusion equation;3D reconstruction3503|214|0更新时间：2024-05-07
- 摘要：A new procedure for reconstructing a smooth parametric surfaces using T-Splines from a triangulation mesh was present in this paper.In our solution,a key ingredient is that the scheme for automatically extract a quad-dominant control mesh(T-mesh)and a parameterization of the data points over the T-mesh.We use the discrete conformal parameterization as the solution of choice for mapping the 3D triangular mesh to a 2D domain,and we partition this two dimensional space by recursively subdividing it into four quadrants,then automatically constructing the T-mesh that we need for the reconstructing.By using least square approximation we can get the control point of the surface and finish our algorithm until the approximation error below a specified threshold.We use adaptive refinement of the T-mesh in order to satisfy user-specified error tolerances and demonstrate our method on real data.关键词：T-spline;T-mesh;fitting;surface reconstruction3145|277|0更新时间：2024-05-07
- 摘要：In order to optimize 3D display performance for volume data of industrial CT,it is necessary to accelerate slice display. Programming method based on graphics processing unit(GPU)is applied to volume display algorithm.Because GPU render complex scenes using a programmable parallel pipeline and slice display algorithm is implemented by graphics application which call fragment shading program, many programmable fragment processors can contemporaneously sample for lots of pixels. A slice display algorithm of CT volume based on GPU is presented. Firstly loading and preprocessing industrial CT volume data are interpreted. Then design steps of Cg application and fragment shading program are presented. Experiments of cutting muli-slice industrial CT images is carried out,the experimental results show that slice image generated by GPU algorithm is clear, the frame rate is steady and high, speed of computing is 2~9 times than CPU algorithm’s.It meets the needs of industrial CT image system in terms of rapid 3D display.关键词：industrial CT;graphics processing unit（GPU);slice display;hardware acceleration2776|231|0更新时间：2024-05-07
- 摘要：With tracing region enlarged, an augmented reality system, which uses magnetic force tracker to realize tracing, need more magnetic force trackers. But their transmitters are very close to each other, it leads to mutual interference and virtual object in helmet vibrating obviously. According to the moving characteristics of head and hand while interacting, different methods are designed and applied to control the mutual interference. To the interference of tracing hand, particle filter is adopted. To the interference of tracing head, Kalman filter is used only when it holds still. While moving slowly, the improved intrinsic 3D discreet Curve smoothing algorithm is used. Compared with old algorithm, this improved algorithm can deal with dynamic data using an approaching-real-time model and remove of interference to Euler Angle using bilateral filtering. While head moves quickly, filter is stopped. Furthermore, the critical damping is applied to avoid jumping while entering the quick state. The results show that the mutual interference is reduced significantly by using these methods.关键词：mutual interference;Kalman filter;particle filter;curve smoothing;critical damping;augment reality2454|158|0更新时间：2024-05-07
- 摘要：A visual robust augmented reality (AR) register algorithm based on retrieving and tracking pre-defined image feature points is given in this paper. Feature points are retrieved from video by voting from random tree. False matching points are removed by semi local constraint paradigm with iterative homograph constraint. Optical flow is used to track the pre-defined tracking points. Rotation and translation matrix from world coordinate to camera coordinate are obtained with coplanar POSIT algorithm. For tracking method being used in our algorithm, dithering problem is effectively solved in real-time experience.关键词：augmented reality;random tree;optical flow2836|189|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