最新刊期
卷 16 ， 期 6 ， 2011
- 摘要：Contour extraction is one of the typical problems in image processing and computer vision. Contour grouping, which is a contour extraction technique based on perceptual organization, shows promising development trend. This paper addresses the research on contour organization. First, it puts forward the relationship between contour extraction and perceptual organization. Second, a comprehensive summary on related research work is given. Then their characteristics and shortcomings are analyzed. At last, further research directions are suggested.关键词：contour grouping;perceptual organization;gestalt law;contour extraction;perceptual3278|234|0更新时间：2024-05-07
- 摘要：A novel watermarking method to construct watermark synchronism by using image specific edge and Directionlet transform was proposed. The direction of the image specific edge is selected as the transform direction of Directionlet, the alignment direction of Directionlet is achieved by forming an invariable angle with the transform direction. The generator matrix of the lattice is given by the two direction vectors.Several copies of watermark are adaptively inserted into coefficients of the selected middle and low frequency subband of all the cost of the lattice. In order to eliminate the effect of geometric attacks，the watermark is detected in the reference frame established by utilizing the direction of the image specific edge as direction of its an coordinate axis. Experimental results show the proposed watermarking algorithm outperforms the Wavelet and Contourlet-based in terms of transparency as well as robustness,especially at the geometric attacks such as rotation.关键词：Directionlet transform;specific edge;transform direction;alignment direction;lattice;coset2807|240|0更新时间：2024-05-07
- 摘要：To reduce the complexity of AVS-M sub-pixel motion vector search，this paper proposes a sub-pixel motion estimation fast searching algorithm based on prediction．The algorithm expels the less possibility points by doing matching error relation of sub-pixel points selected analysis, which assures the accuracy of sub-pixel search. Through effective search strategy, compared with the half-pixel full search algorithm and CBFPS, experiments show that the proposed algorithm can respectively reduce the sub-pixel search points by 65.25％87.69％ and 43.67％67.38％ on average PSNR lost less than 0.01dB to the video sequences with different motion characteristics. The algorithm efficiently decreases the computational number of sub-pixel motion estimation．关键词：AVS-M;video coding;sub-pixel motion estimation;prediction vector3200|200|0更新时间：2024-05-07
- 摘要：Motion estimation is the most important and time-consuming part in H.264. A large number of algorithms are based on the reduction of search points to reduce the computational complexity. In video sequence, people are often interested in moving objects. There are many static blocks in background, and the movement between the two different background of inter-frames is often considered as a result of camera moving. According to image feature of background block and foreground block and using MV prediction, we propose a new algorithm. It is based on these strategies, including static block determination, background-block, foreground-block and edge-block detection, start point prediction and the improvement of search window. Comparing with fast-full search algorithm (FFS), UMHex and SUMHex, experimental results reveal that our algorithm can reduce motion estimation time, while the PSNR value is kept approximate to FFS，with a slight increase in bit rate.关键词：motion estimation;background-block;foreground-block;start point prediction3267|219|0更新时间：2024-05-07
- 摘要：In order to increase measurement precision in Profilometry, reduce possible frequency overlaps and tolerate the height discontinuity of complex three-dimensional shape, an improved Profilometry using a grid carrier fringe pattern is presented. In the Profilometry, two spatial fringe patterns are perpendicular to each other to avoid frequency overlaps; two one-dimensional deformed grating patterns are separated from the deformed grid pattern by use of both two-dimensional Gabor transform and variable rectangle bandpass filters, from which the wrapped phases in x and y directions are respectively extracted. The corresponding unwrapped phases are achieved by use of a lookup table method. Compared with Profilometry using single frequency fringe pattern and one-dimensional double-frequency fringe pattern, the improved Profilometry could obtain double phase information of identical precision, increase measurement precision and realize discontinuous phase unwrapping. Detailed theoretical analysis is presented, and the simulated and experimental results show that it is feasible.关键词：information optics;profilometry;two dimensional Gabor transform;two dimensional grid grating pattern3143|223|0更新时间：2024-05-07
- 摘要：Suppose each pixel of an image is a random variable under some kind of probability distribution, according to the Bayes theorem, the segmentation of the original images is equivalent to their maximum a posteriori probability estimation. In this framework, we proposed an improved image segmentation algorithm based on Graph Cuts. The construction of the original Graph Cuts model is improved in two aspects. First, fuzzy C-means clustering is introduced into the energy function of data restriction. With the help of fuzzy clustering method, the energy function’s performance of constringency is improved. Second, nonparametric method is used to estimate the statistical distribution of the image, which work as the prior probability used in image segmentation. With the presented method, the results of segmentation are guaranteed to be smooth locally. Since the nonparametric estimation is directly evaluated from the samples, and is suitable for situations of small samples and variable distribution functions, the applicability of our algorithm is extended. Experimental results have shown that the proposed algorithm has good performance on segmenting remote sensing images and medical images.关键词：Graph Cuts;Bayesian image segmentation;fuzzy C-means;nonparametric estimation3942|303|0更新时间：2024-05-07
- 摘要：This paper proposes a new region-based active contour for existing images segmentation model to light-sensitive. The energy functional consists of a geometric regularization term that penalizes the length of region boundaries and a data fitting term. Particularly, the local entropy is used as the data fitting term to distinguish different region. First, this paper uses a sliding window function to extract the local entropy according to the relationship of spatial arrangements of image pixel, which can map intensity space of image to local entropy space. Then, we can get the region competition model by maximum a posteriori segmentation probability in local entropy space. Next, it has a fuzzy region competition model by the membership function to replace the characteristic function to solving this model. Finally one can solve this model using fast Chambolle’s dual method. The experimental results for some images show desirable performances of this model, which has the fast convergence speed and light stability.关键词：image segmentation;active contours;variational;fuzzy;local entropy3772|282|0更新时间：2024-05-07
- 摘要：Weight calculating formulas of existing threshold segmentation algorithms based on graph spectral theory via normalized cut do not pay enough attention to the relationship between pixels, can not get the real solution when images have weak edges and thus cannot segment the details of images very well.The proposed algorithm pay enough attention to the relationship between pixels when calculate weight by introducing a new constraint which is made by Gaussian Mixture Model to the algorithm. Before computing normalized graph cut measure, proposed algorithm computes the distribution of threshold range adaptively by the median parameter of Gaussian Mixture Model, therefore the proposed algorithm makes the computation of normalized graph cut measure very efficient. Experiments show that our algorithm performs better in segmentation and preserve more details than existing threshold segmentation algorithms based on graph spectral theory via normalized graph cut measure.关键词：graph spectral theory;image threshold;Gaussian mixture model;normalized graph cut measure2941|230|0更新时间：2024-05-07
- 摘要：A new method based on C-V active contour model is proposed to extract phase-change line automatically from complex phase-change thermography sequence. First, an accurate image matching measure based on structural characteristics is developed to solve the location bias in some frames of phase-change thermography sequence. Second，effective phase-change information is obtained according to the inter-frame difference between the template image and the differential image contrast enhancement. Finally, a phase-change line tracking algorithm is presented by introducing shape priori characteristics into C-V active contour model. The method is able to rapidly solve multi-phase change lines and topology change problems which can’t be solved with existing methods. The effectiveness of the proposed algorithm is validated and compared to alternative methods with variant test models. The software system based on the algorithm has been successfully applied in practice.关键词：phase-change thermography;C-V active contour model;image segmentation;image sequence2927|365|0更新时间：2024-05-07
- 摘要：Influenced by appearance changes and occlusions, template updating has been one of the most difficult problems in the object tracking algorithms based on pattern matching. Because both appearance changes and occlusions can lead to the changes in the intensity of object, however, changes that result from the obstacles can’t be updated as the template; while changes that result from the object in itself should be updated in time. Therefore, a local template update method with the discrimination between occlusions and appearance changes is presented in this paper. The proposed method uses the fragment template for the tracking and distinguishes occlusions from appearance changes according to the probabilities of which varied gray level information belongs to the target and to the background region. The template is preserved to prevent the information of the obstacles from contaminating the template when the target has been partially occluded; while a new adaptive local template update algorithm is used in the case of appearance changes. Experimental results are presented to demonstrate that our algorithm can be adaptive to the appearance changes as well as occlusions and is more robust than the total model update strategy.关键词：fragment tracking;template update;appearance change;occlusion4095|243|0更新时间：2024-05-07
- 摘要：A deficit of classic Gaussian mixture model in background subtraction is the high computation cost. To solve this problem, a novel algorithm is proposed in this paper. A threshold parameter corresponding to the mean of deviation is utilized to judge whether a model matches the current pixel intensity. The new algorithm efficiently reduces the calculation burden of the operation of square and exponent with classical model; meanwhile, a non-linear weight updating approach is adopted, with the notion of sustain stationary time introduced in, and hence the quick converting of a relative long still object in scene into the background is achievable. Experimental results show that the new algorithm has significantly improved the calculation efficiency of background model.关键词：object detection;background model;Gaussian mixture model;weight updating2957|272|0更新时间：2024-05-07
- 摘要：Text detection plays an important role in the system of text information extraction. To solve the problem of detecting text in the bad settings such as complicated Text-background, uneven illumination and so on, a generation text detection method based on stroke direction map is proposed, which utilizes a coarse-to-fine framework. In the coarse step, it firstly builds the stroke direction Information map of the source images with Haar wavelet and LBP descriptor. Then the candidate text lines can be obtained through filtering, component analysis and projection profile analysis of the direction Information map. In the final step, it makes use of SVM classifier with kinds of text texture features to verify the real text lines. Experiment results on ICDAR03 show that the proposed algorithm can quickly and accurately detect texts region in different settings.关键词：text detection;Haar wavelet;LBP;SVM3578|229|0更新时间：2024-05-07
- 摘要：As a fundamental unit in video analysis, automatic shot detection and classification plays a significant role. To keep consistent with the characteristics of human visual perception, the semi-supervised label propagation based shot boundary detection and classification technique is proposed in this paper. Taking the correlations among consecutive frames in video stream into consideration, the pre-constructed initial state of label for each shot category is propagated continuously via correlation graph，of which the final convergent state can be exploited to reveal the intrinsic description of various shot categories. Furthermore, we apply a multi-class SVM to fulfill the shot classification. The experimental results show the effectiveness of the proposed algorithm, from which the performance of video analysis and retrieval can be expected to benefit.关键词：shot detection;label propagation;shot classification;support vector machine3188|187|0更新时间：2024-05-07
- 摘要：Moving objects extraction is a key part of video surveillance system. To improve the performance of moving objects detection method based on the Gaussian Mixture Model(GMM), an iterative detection algorithm with adaptive partitioning block of pixels is proposed. It is based on the temporal-spatial background that the number of components is improved adaptively and the feature of areas extracted spatially is combined. With the spatial areas information, the algorithm decreases the number of small fake objects and reduces the fragmentation of objects that caused by all kinds of noise. Comparing with detection method based on single pixel, the proposed method would not almost impact the detected results when it reduces the algorithm computation obviously. The results show that the objects extracted by the proposed method with higher SNR and the processing time decreases 22% contrasting to traditional algorithm.关键词：GMM;areas of background;adaptive;partition block detection2839|302|0更新时间：2024-05-07
- 摘要：Animating human bodies becomes one of the most active research fields in virtual human modeling. Animating an articulated three dimensional human model currently requires manual rigging to specify its internal skeletal structure and to define how the input motion deforms its surface. Manual rigging is too tedious, and sensitive to the pose of the model. So in this paper, we present an automatic pose-independent method to extract skeleton of human model. This method based on geodesic distance of human body model first extracts and recognizes five feature points at the end points of head and limbs on human body surface, and calculates the geodesic distances of vertices in the model from the feature points. It then constructs the central lines through the centers of geodesic distance isolines, and finds joint positions by analyzing the angles or circularity function of geodesic isolines of the model according to bent or unbent joints. Our experiments show that, this method can identify joint positions and extract the skeleton automatically, with high precision, and in a pose-independent manner.关键词：virtual human body model;pose-independent;geodesic distance;skeleton extraction3886|733|0更新时间：2024-05-07
- 摘要：Fingerprint image thinning is an essential step in the fingerprint recognition system, which plays a key role in the feature extraction. In order to thin the fingerprint image efficiently, we analyze the underlying causes of the disadvantages of the existing fast parallel thinning algorithm and the improved OPTA thinning algorithm, and then present a new composite fingerprint image thinning algorithm is presented in this paper. A set of preprocessing templates are designed to smooth images after fast thinning. In addition, the improved OPTA algorithm is optimized for speed. A large number of experimental results show that, the presented algorithm has high thinning speed, and can obtain good thinning quality.关键词：fingerprint;thinning;parallel algorithm;template3856|263|0更新时间：2024-05-07
- 摘要：Modified Multi-valued Exponential Correlation Associative Memory Model (MMECAM) is a neural network with higher storage capacity. In this paper, based on the analyses of the strengths and weaknesses of MMECAM, a new Gauss Auto-associative Memory Model (GAM) is proposed by modifying its update rule. Then the stability of the proposed GAM is tested in synchronous and asynchronous update modes with a defined energy function, which ensures that the learnt patterns become stable points of the GAM. Further, a framework of Generalized GAM models (G-GAMs) is presented by introducing general similarity measures which makes GAM become its special cases. Finally, the GAM is applied to image recognition from a single sample per image successfully, and the computer simulation results verify GAM’s robust performance.关键词：associative memory;stability;similarity measure;neural network;image recognition3825|188|0更新时间：2024-05-07
- 摘要：The intensity change of the image quality (IQ) can reflect the photography condition change of a CCD camera. A novel air quality evaluation method, which only uses the surveillance image, is presented in this paper. First, different IQ metrics, i.e. the noise brightness ratio, the contrast level and the blur level, are employed to evaluate the IQ. Then we use them to build different time series data. After that we utilize a Gompertz type diffusion process (GTDP) of stochastic growth and the detrend fluctuation analysis (DFA) method to analyze and fit the trend of these series above. Finally, we can use the parameters of the GTDP model and the DFA model to evaluate and classify the IQ change level of different atmospheres. Some experiment data, which are captured in a solar day and a misty day, are utilized to verify the validity of our method.关键词：image quality evaluation;DFA;diffusion process;air quality evaluation3361|189|0更新时间：2024-05-07
- 摘要：This paper investigates automatic construction of a three-dimensional (3D) heart model from 4D medical images. It provides reliable functional analysis of human hearts for diagnosis and treatment for cardiovascular diseases. Firstly, a 4D image is constructed from a set of medical images that are obtained at a sequence of time points. Then, based on a statistical shape model trained from thousands of patient image sets, a geometrical and statistical heart model is generated. With this model, for a new case, an adaptive segmentation and fitting scheme is used to obtain its 3D flexible shape which is interpreted as the deformable surfaces of the patient heart. Finally, functional evaluation of the cardiac ventricles can be followed by quantitative analysis of some important parameters, such as volumetric measurements, myocardial dynamics, and disease analysis. Experiments are carried out to demonstrate the proposed method. The study provides an important means of diagnosis and treatment for heart diseases.关键词：heart;4D image;ASM;flexible model;image processing3406|246|0更新时间：2024-05-07
- 摘要：This paper presents a novel method for recognizing object categories. In this method, matting technique is applied to estimate foreground distribution information, which provides weights for image blocks. Blocks describing foreground are given higher weights, while blocks describing background are given lower weights, then all the blocks are sorted according to the assigned weight.We also conducted imagery comparison study. First of all, distances between blocks with corresponding weights are evaluated, and then the similarity between images can be obtained through adding all the distances between blocks by weights. Experiments on challenging object categorization image sets demonstrate the effectiveness of this method.关键词：image retrieval;matting;spatial cues;image blocks2694|215|0更新时间：2024-05-07
- 摘要：The performance of image matching based on S-dimensional assignment algorithm and K-means clustering algorithm are both poor when the feature points within the images are severely missed. Besides, the computational complexity of the aforementioned two algorithms increases dramatically when the number of images to be matched increases. Aiming at these problems, a new decision algorithm for image matching decision is proposed which uses the truncation of the feature point sequence. The algorithm takes advantage of the prior knowledge that the matching measures between the feature points which are matched are larger than those which between the feature points which are not matched. So on the one hand it has good robustness to feature points missing problem, on the other hand it overcomes the combinatorial explosion problem in the aforementioned two algorithms. Simulations validate the effectiveness of the proposed algorithm.关键词：robot vision;image matching;S-dimensional assignment algorithm;trunction of feature point sequence2904|208|0更新时间：2024-05-07
- 摘要：The 3D models reconstructed from binocular stereo always have noticeable noise, and even contain holes. To deal with this issue, this paper presents a non-local (NL) method to denoise and hole-fill the original model on a regular vertex set. To improve computational efficiency, the proposed method uses a basic NL method and principal component analysis (PCA) based NL method to remove model noise in hole and non-hole regions, respectively. The filled holes are then processed by NL denoising to improve the consistency with the surrounding surfaces. The experimental results show that, when evaluated on different real objects with various sizes, the proposed method can reconstruct 3D models effectively and performs much better than previous methods.关键词：binocular stereo;non-local;denoising;hole filling;principal component analysis3553|226|0更新时间：2024-05-07
- 摘要：This paper presents an enhanced IBFV visualization algorithm based on particle advection, it can obviously increase the rendering image contrast. The particle advection is at first performed to get a series of vector textures. Then the vector textures are performed image blending with texture of frame-buffer to generate rendering images, which are treated as the background images of IBFV algorithm instead of noise textures. In this way, it can not only exactly reflect flow dynamic variety, but also enhance the contrast among flow lines. Furthermore, this algorithm can achieve a real-time frame rates as fast as IBFV algorithm.关键词：vector field visualization;particle advection;texture mapping;image blending3634|210|0更新时间：2024-05-07
- 摘要：Import motion capture data which have the matching skeleton with character is often used in preset scene in character animation synthesis now. However, this method could not afford the demand of multiple topological skeleton structures and the changing scene in real time. In view of the above questions, this paper proposes a new algorithm of retargeting motion capture data to skeleton model with multiple arbitrary topologies and adopts intelligent character model based on the real time three dimensional path-planning algorithm, integrating with the method of using voice user interface instead of graphical user interface to implement verity movement of virtue character in a dynamical three dimensional scene. The result demonstrates that slick and interactive animations can be synthesized by the proposed method. The method downs the cost of animation synthesis by improving the reuse rate of motion data and can be used widely in human-computer interaction domain by satisfying the need of synthesis of character animation in varied three dimensional virtual scenes.关键词：motion capture;motion retargeting;intelligent path-planning;interaction between animation and scene;speech recognition3125|266|0更新时间：2024-05-07
- 摘要：The noise analysis evaluation and filtering of remote sensing image have always been an important research field.In recent years,nonlinear diffusion model based image de-noising have received considerable attention in the field of remote sensing image application since the model can efficiently remove noises in remote sensed images,while well preserving the texture information.The paper proposes a new nonlinear diffusion model by introducing wavelet modulus maximum into the diffusion model and gives a discrete scheme.Our model removes the noise better than P-M model. Meanwhile,the proposed model overcomes the shortage of ALM model that tend to bluring and losing singular point. Our model can not only efficiently remove noise in remote sensing images,but also simultaneously retain detail information, such as edge and texture.Experimental results illustrate the effectiveness and stability of the proposed model.关键词：remote sensing imge;de-noising;wavelet transform maximum;nonlinear diffusion2894|237|0更新时间：2024-05-07
- 摘要：In this paper, we propose a concept of corner measurement according to Harris corner detection principle, and take corner measure amplitude as a choice of fused coefficients for high-frequency image. And then the significant central coefficient (SCC) image fusion algorithm based on corner measure is presented in redundant wavelet field. The basic idea of the proposed algorithm is first to decompose an image into wavelet planes and approximate planes by using redundant wavelet transform, and then use the corner measurement response function to evaluate corner measure amplitude for the wavelet plane. Significant central coefficient (SCC) image fusion algorithm based on corner measure fusion rule is adopted for wavelet planes. For the approximate planes, average coefficient fusion rule is adopted. Finally, the fusion image was obtained by taking redundant wavelet inverse transform. In the experiment, we adopt multi-spectral data of Clementine lunar surface and multi-spectral data of SPOT5 to verify the validity of our algorithm, and we also compare it with other methods. Beside the subjective comparison based on vision, we introduce objective evaluation index such as the standard deviation, entropy, clarity and the correlation coefficient of the integration to evaluate fusion results. The experiments show that the algorithm can maintain details information feature such as the corner and edge of source images more effectively.关键词：image fusion;corner measure;redundant wavelet transformation;multi-spectral image of lunar-surface3380|248|0更新时间：2024-05-07
- 摘要：High radiometric resolution (≥10bit) digital remote sensing image obtained by aeronautics and astronautics technology has gradually been the main data source of spatial geographic information. Focusing on the characteristic of high radiometric resolution digital remote sensing image, this paper presents a reversible information hiding algorithm based on integer DCT(discrete cosine transform), which uses a high frequency coefficient modification technique in the integer DCT domain to modulate the information, the corresponding embedding capacity increase scheme and distortion control scheme are also proposed. During the information extraction process, the original image can be perfectly restored with the embedded information accurately extracted at the same time. The results of experiment proved that this algorithm is invisible, can efficiently against linear stretching attack, and has better robustness to crop, rotation, noise or filter attack.关键词：remote sensing image;high radiometric resolution;reversible information hiding;integer DCT3024|201|0更新时间：2024-05-07
- 摘要：To facilitate designers to design 3D face shapes conveniently, this paper presents a method of 3D face modeling using freehand portrait sketch. Our method has two characteristics: Firstly, pose estimation is introduced to face sketch analysis. It infers the corresponding front-view sketch from the side-view face sketch, users then can draw the free-form portrait sketch from their selected view. Secondly, multi-layer mapping is adopted to establish one-to-one correspondence between the sketch feature points and 3D face feature points, and the displacements between the corresponding feature points are calculated to generate the specified 3D face models. It guarantees that the geometrical characteristics of the strokes can be retained in the final 3D face model. The experiment results prove that the proposed method can generate novel 3D face models quickly and support the process of sketch-based face modeling efficiently.关键词：sketch-based interface;3D face modeling;pose estimation;feature point extraction;mesh deformation4303|284|0更新时间：2024-05-07
- 摘要：This paper presents a novel physically-based parallel approach to simulate tree motion using GPU in real-time. The tree’s kinematic principle and the hierarchical matrix structure model (namely HMSM)’s parallel mechanism are analyzed to realize the speedup of the trees animation on CUDA. Firstly, we briefly introduce a method of physically-based tree motion called hierarchical matrix structure model (namely HMSM) driven by the external forces such as wind; then we analyze this model in a parallel management theoretically in detail; thereafter, describe the design of parallel structure and parallel algorithm of the tree’s motion, which are suitable to implement on CUDA;finally, realize the physically based motion simulation on GPU. Experimental results show that the presented approach can animate trees realistically and naturally, and simulate physically based trees motion in real-time. Moreover, the approach give good clues for speeding up the algorithm for the computer animation.关键词：tree animation;physically-based;hierarchical structure;CUDA accelerate;real-time3706|214|0更新时间：2024-05-07
- 摘要：Feature visualization plays an important role in visualization of complicated flows because it can highlight the feature of the flows with a simplified representation. Traditional feature visualization methods usually employ the semi-automatic technique. The user’s interactive requirements can hardly to be met due to the weakness for integrating the user’s knowledge and experience. Because of the difficulties with extracting the strict confines of a feature, this paper presents an interactive fuzzy feature extraction algorithm, and has realized interactive extraction and visualization of the flow features. Firstly, construct a model describing the flow features based on the fuzzy theory. Secondly, we implement the algorithm based on the syntax standard. We have developed a visualization system based on the presented interactive fuzzy feature extraction method. The experimental results show that our method can not only make full use of the knowledge and experience of the user to extract the features of the data and the objects on demand but also reflect the uncertainty nature of the data.关键词：interactive fuzzy feature extraction algorithm;human-computer interaction technology;interactive visual analysis3153|192|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