最新刊期
卷 10 ， 期 9 ， 2005
- 摘要：Segmentation becomes a difficult task when the background illumination changes.In this paper,we apply a Bayesian learning method into video segmentation.The constantly changing background has been modeled at the pixel level.The feature vector for each pixel is represented with a discrete probability distribution function.The histogram colors and co-occurrence vectors have been calculated.Bayesian learning has been used to obtain these probability distribution functions from the video image inputs.The experimental results indicate that the proposed approach is able to learn a complex background of which the illumination changes either gradually or suddenly.关键词：video image segmentation;bayesian learning;complex background modeling2602|183|0更新时间：2024-05-08
- 摘要：In order to solve the problem of the negative effects of shading and uneven illuminations in the color image on segmentation,this paper proposes a novel algorithm for color image segmentation based on vector quantization and region growing.The algorithm takes into account both color information and spatial information of color images,which avoids the insufficiency of previous methods which have considered either the color information of the image,or the spatial information of the image.Firstly,the color image is quantized by using a modified generalized Lloyd algorithm(GLA),and then selects seed pixels in terms of the quantization results of color images.Secondly,from each seed pixel region growing is performed based on the vector angle similarity criterion.Finally,fuzzy C-means(FCM) algorithm is used to classify the remaining pixels which have not been classified into regions being formed earlier.The experiments show that the negative effects of shading and uneven illuminations in the color image on segmentation can be overcome to a great extent by using the proposed algorithm.Color image segmentation results of the proposed approach hold favorable consistency in terms of human visual perception.关键词：color image segmentation;vector quantization;seed determination;region growing3034|258|0更新时间：2024-05-08
- 摘要：A soft image segmentation algorithm based on the generalized fuzzy set is presented in this paper.We incorporate Gibbs random field into the generalized fuzzy set to compensate for the spatial information,and a generalized fuzzy Gibbs random field model is proposed,and the generalized fuzzy Gibbs segmentation algorithm(GFGS) is developed.Each class is considered as a generalized fuzzy class,and the segmented image is regarded as a generalized fuzzy set on the label set in the proposed algorithm.With the proposed algorithm,the outliers in the image data are described by the negative part in the generalized fuzzy membership function,and can be dealt with effectively.Maximum a posteriori(MAP) is used as the statistical segmentation criteria,in which the generalized fuzzy Gibbs random field is used to obtain priori knowledge.Every class center is updated by the centroid of the generalized fuzzy class.Experimental results on both MR real data and the stimulated brain data show that the proposed algorithm is robust,which can filter the noise and partial volume effect significantly.关键词：generalized fuzzy set;generalized fuzzy Gibbs random field;image segmentation;partial volume effect2928|249|0更新时间：2024-05-08
- 摘要：In this paper,a two-stage search fast algorithm for adaptive projective decomposition,which is used in image segmentation,is presented.In the proposed approach,the first stage is to find the approximate values of the mean and the standard deviation of the Gaussian element in image histogram.And the second stage is to search the exact mean value by 0.618 algorithm.Having the exact mean,some projection values are computed by projecting the histogram to some appropriate Gaussian basis functions,and then the standard deviation and the power can be estimated well based on the least-square curve fitting analysis.Finally,the optimal thresholds between different regions can be determined.The theoretical analysis and the comparison experiment show that the proposed approach is more efficient and effective.关键词：image segmentation;adaptive orthogonal projective decomposition;fast algorithm2865|237|0更新时间：2024-05-08
- 摘要：On the base of combining change-detection-based segmentation approach and spatial edge information by canny edge detection,an algorithm is proposed in which local contrast enhancement is applied to improve the contrast between foreground object and background in the pre-processing stage,and the problem caused by low contrast is solved.A filter was designed to remove a small quantity of noise caused by contrast enhancement;Then for the complex background,the algorithm utilizes probability-based classification to accumulate the background information,which it is needed by the original segmentation algorithm,and consequently realizes the capturing of background information automatically; Finally,the paper proposed that three situation should be discussed in the process of accumulating background information.The proposed algorithm is evaluated on several MPEG- 4 test sequences and produces promising results.关键词：video segmentation;canny edge detection;contrast strengthen;probability-based classification2488|163|0更新时间：2024-05-08
- 摘要：A new efficient approach for the removal of impulse noise is presented based on the characteristic of impulse noise in this paper.The algorithm is composed of two steps.The first step is to identify the impulse noise of image.The following step is to smooth the identified noise pixel,where we take into account the impulse noise characteristic and utilize the topological connectedness of impulse noise.In the second step,using the information around the noise pixel restores the noise pixel.Simulation results demonstrate that the proposed algorithm is obviously better than traditional median-based filters and switching median filter,and is particularly effective for the cases where the images are very highly corrupted.关键词：filters;impulse noise;topological connectedness2631|194|0更新时间：2024-05-08
- 摘要：Point-clouds registration is widely applied in reverse-engineering,computer vision and medical imaging,etc.Based on the combined transformation theory of the space coordinates,a registration method,which extracts the transition matrix with multi-points based on least-square method to minimize the transition error,is proposed.Some typical methods are specified and analyzed with practical examples in this paper.Experimental results show that this approach can realize point-clouds registration as the features of rapid speed,high accuracy,stability and simple operation.It is powerful for practical engineering.3109|244|0更新时间：2024-05-08
- 摘要：For image texture analysis,an invariant texture recognition algorithm is proposed based on Radon transform.Firstly,Radon transform is used to project the image to 1-D space,and then the projection data is transformed via a translation and scaling invariant adaptive 1-D wavelet transform,thus the feature matrix with translation and scaling invariance is derived.Multiscale analysis is employed for the feature matrix,and the energy values at different scales are proven not only to be invariant under image translation,scaling and rotation,but also to reflect the different energy distribution of the texture image at different scales.After the feature vector is availabe,support vector machine(SVM) is used for classification.Comparing with other methods simulations are given to obtain insight into the effectiveness of our method.关键词：RADON transform;invariant;multiscale;support vector machine(SVM)2833|259|0更新时间：2024-05-08
- 关键词：DEM;topographic correction;satellite remote-sensing imagery3124|211|0更新时间：2024-05-08
- 摘要：Motion artifact suppression is a very difficult problem in magnetic resonance imaging.Patient motion including physiological motion and physical motion causes the phase distortion in the collected signals and induces motion artifacts in the reconstructed image for the 2-dimension Fourier Transform imaging method.As the result of these motion artifacts,the quality of image is degraded and the precise orientation to the focus in clinic is affected.A motion artifact reduction method based on genetic algorithm is presented in this paper.Genetic algorithm has the characteristics of parallel,randomicity and adaptive matching pursuit.In the image reconstruction,before taking the inverse Fourier Transform,the phase distortion of K-space signals is compensated step by step through searching for the optimizing phase values.The experiments show that the phase distortion can be estimated using the information implied by the motion artifacts and a significant amount of motion artifact suppression is achieved.Using this algorithm,the corrected image is satisfied when motion is slight,and the quality of the image is still improved greatly in the conditions of noise and significant motion comparing with using classical iterative algorithms.关键词：magnetic resonance imaging;motion artifact;genetic algorithm4479|203|0更新时间：2024-05-08
- 摘要：Labeling algorithms have got broad applications for shortest path finding in transportation networks,among which various fine-tunned Dijkstra's algorithms well known as typical label setting algorithms have been selected by many GIS related software for network analysis.However,label correcting algorithms,the other group in label algorithms family,are rarely used yet in GIS network analysis.After detailed discussion on the structures of labeling algorithms,in this paper,the implementation,complexity and applicability of labeling setting and label correcting algorithms are analyzed.Then three best-known fastest label algorithms,i.e.,Dijkstra algorithm implemented with approximate buckets(DIKBA),Dijkstra's algorithm based on quad-heap priority queue(DIKQH) and Pallottino algorithm(TWO_-Q),were used to carry out practical evaluation on three real urban road networks.The results showed that for one-to-one shortest path calculation,DIKQH and DIKBA greatly outperformed than TWO-Q algorithm,and DIKQH exhibited the best running efficiency.For one-to-all shortest path calculation,however,TWO-Q algorithm runs a little faster than DIKQH and DIKBA on the selected real road networks.The author argued that more attention should be paid on TWO-Q algorithm for its efficiency and applicability.关键词：shortest path algorithms;label algorithms;complexity;transportation networks3092|202|0更新时间：2024-05-08
- 摘要：Path Planning for mobile robots is one of the most fundamental and complex problem in robotics.Main solution methods for path planning mainly include potential field,unit decomposing and neural network methods.Level set methods have been used in a variety of image processing and computer vision tasks with many advantages such as handling of topological changes,numerical stability and independence of paramerization.In order to exploit the application of Level set method in robot path planning problem.Based on introduction of the basic principle,some relative technology of level set method,and solution methods for path planning,implicit Snake or level set model for path planning problems is presented,and fast marching method(FMM) is used to solve this kind of model. Some computation results and their visualization interfaces for this model are given,and compared to the results from classical potential energy methods.Theory and computation results prove that Level set method for robot path planning is feasible and valid,then new technology and method are provide for robot path planning research.关键词：Level set method;path planning;robot vision;Snake3143|187|0更新时间：2024-05-08
- 摘要：A new solution for outdoor augmented reality is introduced in this paper and its architecture and principle are described.Under the condition that the depth of field is much shorter than the distance between camera and scene,a registration method based on panoramic-video information is presented,which is an FFT-based image matching method.The FFT-based method will work to match the camera-captured image with the pre-treatment image,and then according to the match information,the virtual model will be transformed and merged to the real scene.Finally the merged scene will be sent to Head-Mounted-Display to achieve augmented reality.Compared to the existing registration method,the proposed method possesses the advantages of low computing-cost,high accuracy and high speed.Experimental results show the correctness of theoretical analysis and the feasibility of its application in the digital reconstruction of Yuanmingyuan garden.关键词：aumented) reality(AR);outdoor AR;image matching;fast Fourier transformation(FFT);environment registration2967|246|0更新时间：2024-05-08
- 摘要：In order to render massive terrain data in real time,this paper presents an improved LOD algorithm.This algorithm firstly simplifies the DEM data by using Mortan coding principle and stores the simplified data with an unfull quadtree.Then,it builds the real-time continuous Lod based on this unfull quadtree according to the relationship between viewer position and grid object space error.The authors bring out a method named "find the neighbors layer by layer" to patch the cracks between different layers and the procedures to find the different type neighbors are presented in this paper.Finally,it wipes off the invisible grids through back-culling algorithm.In order to improve the rendering speed,it uses Hilbert filling curve method to store and interleaved quadtree method to access the unfull quadtree nodes.The authors use the improved method to simulate the terrain of Jingjiang area and obtain a good effect.关键词：level of details(LOD);terrain;Mortan coding;quadtree;back-culling3553|191|0更新时间：2024-05-08
- 摘要：Based on research on the latest literature and the understanding to forest-fire home and abroad,we provide a new kind of model by particles system to real-time simulate and render the large-scale forest fire.In this model,we improve the traditional particle system in the attribute,the movement and the rendering method of individual particle.In addition,we use the hardware-accelerated texture technology to improve the effect and efficiency of rendering.Meanwhile,the technologies adopted in the model also can be applied to other fuzzy objects(such as smoke,cloud and so on).关键词：particles system;forest fire;texture synthesis;real-time modeling2627|269|0更新时间：2024-05-08
- 摘要：The goal of stereo vision is to recover the three-dimensional information of a scene,and the core of stereo vision is to find the corresponding pixels.However,the correspondence search is too time-consuming for high speed robot vehicles autonomous navigation,even if state-of-art general-purpose processors are used to accelerate it.Aiming at this problem,this paper presents a design scheme for binocular stereo vision system based on field programmable gate arrays(FPGA),its hardware structure is introduced,and a fast zero mean sum of squared differences(ZSSD) stereo matching algorithm is discussed.Moreover,a pixel-serial and window-parallel architecture based on FPGA processor is proposed to achieve ZSSD matching algorithm,which is suitable for parallel processing.The stereopsis is captured by two same decoder chips.After rectification and ZSSD matching implemented by FPGA,dense disparity map is computed and sent to general-purpose computer by PCI bus.The proposed scheme is very robust and exhibits great performance,such as high speed.And the hardware system has high stability and reliability.In addition,this scheme is also applicable to fast and real-time processing of other conventional area-based stereo matching algorithms,such as sum of absolute differences(SAD),sum of squared differences(SSD),etc.关键词：binocular stereo vision;fast stereo matching;zero mean sum of squared differences(ZSSD);field programmable gate arrays(FPGA)3961|230|0更新时间：2024-05-08
- 摘要：Restoration of atmospheric turbulence-degraded image is very important in the field of astronomical imaging and astroobservation.It needs to be solved as soon as possible.Solving this problem can deblur the atmospheric turbulence-degraded image and improve the capability of object identification which is good for late stage works such as object feature extraction and recognition.In order to restore the turbulence-degraded image efficiently,a single frame blind deconvolution image restoration algorithm with two circulations based on Bayes theorem is presented.Its fast implementation is studied and some experiments to analyze the stability are carried out.The experimental results show that the algorithm is capable of resisting-noise with robustness;especially it is more suitable for the situation without any prior knowledge.So the algorithm possesses practical value.关键词：atmospheric-turbulence;image restoration;blind deconvolution;point spread function(PSF)3254|221|0更新时间：2024-05-08
- 摘要：A new degraded image restoration algorithm which uses adaptive smoothness constraint based on fuzzy functions is used,is proposed.In order to suppress noise in image,simultaneously,preserve the details during image restoration,the algorithm uses the fuzzy functions to estimate the number and orientation of line elements in each image region,then these edge estimation results lead to Corresponding smoothness constraint.Furthermore,the smoothness constraint is updated along with image restoration iteration,such that the algorithm can suppress noise in flat regions and preserve details of existing edge.The experimental results show that it has faster convergence speed,better objective quality and subjective vision effect.关键词：fuzzy;smoothness constraint;edge-detecting and preserving;image restoration2781|193|0更新时间：2024-05-08
- 摘要：Controposed to single description coding(SDC),multiple description coding(MDC) for image is source coding in which several bit streams(descriptions) of original image are produced such that acceptable quality of the reconstructed image can be obtained from each description.The quality of reconstructed image only depends on the number of descriptions,the more descriptions,the higher quality of reconstructed image is obtained from combining these descriptions.MDC scheme is well suited to the channels which transmission conditions are very poor for image communication without channel coding,especially for networks and wireless channels.In this paper,the theoretical and practical techniques of MDC are exposed,the advances of MDC are described for these years.A MDC scheme of image based on EZW(Embedded Zerotree Wavelet) is presented.The simulation results show that this method is simple and efficient,and implements robust transmission of compressed image considering the compression efficiency.关键词：multiple description coding;image compression;embedded zerotree wavelet5332|195|0更新时间：2024-05-08
- 摘要：We studied the specific requirements on optical wavelet filter in spatial frequency domain for image data compression based on human visual system(HVS) and image perceptual quality evaluation criteria.The upper boundary estimation of the optical wavelet filter in frequency domain was derived and verified by simulation.It will be the bases in designing such kind of optical filter for high quality image data compression.关键词：optical filter;wavelet transform;image data compression;relative error;human visual specificity2604|189|0更新时间：2024-05-08
- 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