最新刊期
卷 12 ， 期 3 ， 2007
- 摘要：This paper presents a comprehensive review and evaluation for color-based skin detection.The literature is reviewed in four categories: colorspace selection,skin color modeling,dynamic modeling and illumination invariance and adaptation.Using a large data set of 1 894 images,we examine whether the colorspace transformation can increase the compactness of skin distribution and the discriminability between skin and nonskin distributions in fourteen 3D colorspaces and fourteen 2D chrominance planes.We also evaluate the classification performance of skin probability map(SPM),Gaussian mixture model(GMM),self organizing map(SOM) and support vector machine(SVM) in said colorspaces.The results reveal that 1) the colorspace transformation cannot improve performance in general,the discriminability and testing performance in RGB and linear colorspaces are better than in other colorspaces,2) the Absence of the luminance component decreases discriminability and performance significantly,3) the performance of Bayes SPM in 3D colorspaces is superior to that of others,4) except 3D Bayes SPM,the 'color preference' of other detector is intrinsic and quite similar,5) the detector using skin and nonskin model simultaneously is better than the detector using skin model alone.关键词：skin detection;colorspace;skin color model;dynamic model;illuminant invariance4201|303|0更新时间：2024-05-07
- 摘要：Genetic algorithm(GA) is a random search and optimization method based on natural selection and genetic mechanism of the living beings.In recent years,GA is used successfully in solving the complex optimization and the industrial engineering problems.Therefore,the research on GA has attracted a lot of attention.This paper discusses and surveys the status and advances in GA research.The basic algorithms,theory,implementation techniques of GA are outlined first.Then many applications of GA in image processing field are reviewed,such as image segmentation and edge detection,image compression,image restoration,image matching,image enhancement,and image reconstruction.At last,several key problems in this field are discussed as well as their development in the future.关键词：genetic algorithm;digital image processing6209|335|0更新时间：2024-05-07
- 摘要：The model-based reconstruction algorithm is adopted in order to conquer the ill-posedness in optical tomography.An edge-preserving,global smoothing generalized Gaussian Markov random field(GGMRF) is imported into optical tomography reconstruction based on the radiative transfer equation in this paper.The iteration optimization method is used to solve the objective function with the GGMRF model by maximizing the a posterior probability,but the gradient computation of the objective function with respect to optical parameters is difficult.Therefore,a novel gradient computation strategy based on gradient tree is proposed.Experimental results show that this kind of reconstruction technology can improve the image reconstruction quality and decrease the ill-posedness compared to the reconstruction algorithm without priori model.关键词：optical tomography;the radiative transfer equation;Markov random field;the adjoint differentiation scheme2846|158|0更新时间：2024-05-07
- 摘要：It is important for 3D Reconstruction of SEM of metal fatigued fracture to study on material fracture which makes quantitatively analysis of the fracture more accurately.As the faultage of metal has excellent fractal characteristic,a new method of calculating SFS(shape from shading) based on neural network model has been proposed and applied to reconstruct the 3D morphology for single SEM of metal fatigued fracture.The experimental results show that the algorithm is effective for the 3D reconstruction of the coarse surface with unknown light direction.关键词：metal faultage;neural network;3D reconstruction;SEM3384|258|0更新时间：2024-05-07
- 摘要：In order to remove the noise efficiently and preserve the sharp features of the models,a denoising algorithm of a robust multilateral filter for point-sampled models is presented.The algorithm takes into account the relationship between noise and underlying geometric information,such as normal and curvature.First,by choosing a control function for a local adaptive optimal neighborhood,the filter window is set in the region with similar normals to avoid the problem of shrinkage and over-smoothing.Second,normals and curvatures of vertices in the optimal neighborhood are estimated by covariance matrix analysis.Third,based on the filter reference plane,normals and positions of surface points are smoothed respectively,i.e.,the normals of surface points are calculated firstly by using multilateral filter,then,by applying multilateral filter again,the position offsets of sampling points are obtained,finally,each point is moved in the direction of normals being smoothed.Experiments show that the multilateral filter can remove the noise efficiently while preserving the geometric features of the surface.关键词：point-sampled model;bilateral filter;trilateral filter;multilateral filter;covariance analysis;smoothing3627|211|0更新时间：2024-05-07
- 摘要：Robust watermarking techniques are always used for copyright protection and ownership assertion.Many watermark embedding schemes in transform domain are proposed such as 2-D DFT,2-D DCT and 2-D DWT which are used to transform the image data,and 3-D DCT and 3-D DWT which are used to transform the video data and 3-D image data.In this paper,a gray-level image watermarking scheme based on 3-D separable wavelet transform(SWT) with lifting scheme is presented.A gray-level image is decomposed into a 3-D image sequences by sub-sample of zigzag scan which are transformed using 3-D SWT.Simultaneously it is proved that the distribution of 3-D SWT coefficients follow the generalized Gaussian density function by using the distribution relative entropy theory.To make the compromise between the robustness and the imperceptivity,a new 3-D HVS model is reconstructed and used to adjust the embedding strength.In watermark detection,the optimum detector is used to implement the blind detection.It is shown in our experiments that the scheme is strongly robust against various attacks.关键词：digital watermark;lifting wavelet;HVS;relative entropy;optimum detection2933|210|0更新时间：2024-05-07
- 摘要：As one of the effective coding tools in the newly-published H.264/AVC coding standard,variable-size block motion estimation is of great importance in performance evaluation of the encoder.A content-based variable-size block partition method is presented for H.264.The proposed algorithm is based on pre-analysis of video content to select between TDVSBM and BUVSBM that best suits the video in unit of the frame.To enable the adaptive selection,a method based on the STGS-image is proposed to develop an efficient and computationally simple criterion.Experimental results show that the proposed content-based variable-size block motion estimation(CBVSBM) algorithm reduces the computational complexity in contrast to both TDVSBM and BUVSBM,but with similar visual quality.关键词：variable-size block ME;adaptive selection;H.264/AVC;STGS-image2695|205|0更新时间：2024-05-07
- 摘要：In order to improve the speed of fractal image coding algorithm,this paper proved an inequality linking the root-mean-square(RMS) and similar measure mathematically and set up two kick-out conditions based on the similar measure and standard deviation of matched block.Therefore it can search out the best-matched block to a range block with a reduced search space at encoding process,which can shorten its runtime significantly.Computer simulation on 4 test images with different complexities demonstrate that the proposed scheme could speedup the coding process,while the subjective quality of the decoded image has a little degradation than that using the corresponding baseline fractal coding algorithm.关键词：fractal image coding;image compression;similar measure2822|173|0更新时间：2024-05-07
- 摘要：This paper presents a new image segmentation and composition method.This method first separates the pixels which obviously belong to objects or background,then gets the cutout whose corresponding value of the new energy function is the minimum.Afterwards the cutout is fused into the target image using gradient comparing method.The gradient comparing method ensures the compliance of illumination in the resultant image meanwhile preserves the color characteristic of the object.It can be easily concluded from the experiments that the new composition method has extended applications and the resultant image has a better lighting than those produced by other methods.关键词：image segmentation;image composition;graph cut method;gradient comparing method2883|193|0更新时间：2024-05-07
- 摘要：In order to extract the moving video object accurately,a novel algorithm for automatic segmentation and tracking moving object is proposed in this paper.Firstly,a fourth-order statistic hypothesis testing in inter-frame difference is used to automatically separate the moving areas from the background in a general video sequence.Then the morphological filter is used to remove the noises and holes.Based on the intersection operation of the two moving areas which obtained from three consecutive frames,the covered/uncovered background may be detected and moving object model can be obtained.Secondly,after determining the initial contour of the moving object,an improved active contour which uses the gradient vector as the external force guides the initial contour moving to the video object contour.Lastly,the motion compensation is made to the video object contour to obtain the initial contour of the next frame and the improved active contour is used again to correct the video object contour of the next frame.Thus,video object tracking can be realized.Our method can not only obtain the accurate contour of the video object but also make the segmentation and tracking for multiple objects.关键词：video object;active contour;hypothesis testing;morphological filter;motion compensation2760|208|0更新时间：2024-05-07
- 摘要：In this paper,we improved the external energy of active contour model,proposed a new adaptive segmentation image model aimed at gray-level image and extended it to the vector value image segmentation.New model couples the fast edge integration and simply statistical methods.The prior information of regions and boundaries of image has been considered sufficiently in this model,and different segmentation models can be constructed based on different probability density function respectively.An instance based Gaussian probability density function has been given in this paper and the AOS scheme that is efficient and unconditional stable has been used to segment the gray image and the vector values image.Compared with Fast Edge Integration method,the experiment results show that the new approach is more accurate and robust and can obtain very good partition.关键词：curve evolution;variational principle;level set;active contour model;log likelihood;additive operator splitting2651|242|0更新时间：2024-05-07
- 摘要：Automatic container code recognition(CCR) plays an important role in custom logistics and transport management.This paper proposes an algorithm for container character location and recognition based on image segmentation and region feature analysis.First,an improved single-pass split-merge image segmentation algorithm is used to segment image based on adaptive threshold,and features of every region are obtained simultaneously.Next,character regions are separated from non-character regions by rules.Finally,model match and neural network are used to recognize every character binary region.Experimental results on 1 804 container images show that the overall accuracy can reach 97.01%(1 750 images) with the proposed algorithms.关键词：image segmentation;character recognition;single pass split and merge;artificial neural network(ANN)3563|185|0更新时间：2024-05-07
- 关键词：the images of intravenous needles;moment invariants;amended I72772|195|0更新时间：2024-05-07
- 摘要：To improve the limitation of Mean Shift lacks method of template update,an algorithm of template update based a group of Kalman filters is proposed.Probability of eigenvalue in feature space is taken as the template information.A group of filters are devised,where each filter is used to estimate the change of probability of sub-eigenvalue.All update value of template can be received by multiplying these corresponding probabilities in sub-feature space.The noise parameter of each filter would change with input data,so a novel strategy of template update according to change of residual of filters could be proposed.Experimental results show that the proposed algorithm can successfully track target under condition of changeable gesture of target and changeable illumination,and is robust to occlusion.关键词：Mean Shift;Kalman filter;template update2974|205|0更新时间：2024-05-07
- 摘要：Face tracking problem comprises two key factors: one is the design of tracker;the other is the selection and representation of tracking cues.To relieve the shortcoming that the general face tracer is frequently disturbed by background when a simple ellipse model is used to describe the facial shape,and in accordance with the fact that the dynamic model and observation model of video object tracking problem are always non-Gaussian and non-linear,a robust particle filter face tracking algorithm which uses shape and color represented in the form of histogram as clues is proposed.The tracking algorithm is based on particle filter,in which a new shape representation method using histogram is introduced and modified to be as one of tracking clues.At the same time,an experiential edge detection algorithm is proposed in order to alleviate background disturbance.Experimental results show that the proposed algorithm is robust to the rotation of face,background skin color distraction and partial occlusion.In addition,the proposed tracker can recover the lost terget due to the severe occlusion.关键词：Face tracking;particle filter;CONDENSATION;shape representation;skin color model2862|244|0更新时间：2024-05-07
- 摘要：Fast and efficient extraction of building models from airborne lidar data is an important research topic in applications of airborne lidar system.The semi-automatic method with user's interaction is a practicable approach for extracting simple and regular buildings from point cloud data.Based on this view,a method of interactively extracting gable-roofed building models from lidar point cloud data is proposed in this paper.An improved Hough transformation in 3D space and Cluster Analysis Model are key points of this approach.The proposed method consists of three steps.First,the user roughly input a region where a building is located.The candidate points belonging to the building roof are then filtered out.Second,an improved Hough transformation in 3D space and Cluster Analysis Model are applied to the data set of roof points to acquire parametric expressions of roof planes.The third step is to construct the integrated building model.The roof's ridge line is obtained by intersecting roof planes,and the building's footprint is determined by analysis of roof points.Finally,a complete building model is created by adding vertical walls.Tested with sample lidar data from Optech Company,this approach gets good experimental results with very few user interactions needed.关键词：airborne lidar;building model;Hough transformation;cluster analysis2600|222|0更新时间：2024-05-07
- 摘要：Object recognition based on contour is a challenging problem in the field of pattern recognition and computer vision.This paper proposes a new attractive method that can be used in this field.Object features based on contour,called Gaussian descriptors,are introduced and their applications to object shape recognition are provided.The method includes defining a Gaussian Potential Function(GPF) and the reafter constructing 8 Gaussian descriptors by calculating the averages of GPF's along 8 circles with the same center.Gaussian descriptors are invariant to translation,rotation,scaling changes and reflection.Moreover,compared with the existing shape features,they are more robust against noise and slight edge variations,have lower computation complexity and higher recognition/retrieval rate,and are application-independent.These properties and advantages of Gaussian descriptors are studied and discussed,and numerical experiments are implemented to show encouraging results and illustrate the promising performance of Gaussian descriptors.关键词：shape feature;gauss potential function;gaussian descriptors3056|206|0更新时间：2024-05-07
- 关键词：liquid crystal display(LCD);colorimetric characterization;S shape model;gain-offset-gamma(GOG) model3515|173|0更新时间：2024-05-07
- 摘要：In order to employ remote control on logistics transportation devices automatically the image controlling molds first collect and process positions of conveyance device.The collected position information is transmitted to the host by the wireless communication mold.A strategy artificial intelligent algorithm control the device's trajectory automatically and stop it at the required position.When the device meet the obstacle while moving forward,it can stop and alarm automatically through the image and optical electrical sensors which perform dual identify.After the obstacle is removed,the device can run along required trajectory.The function of the device can assure the flexible requirement of manufacture system.It's been proved by running experiment that the functions can be realized successfully.The logistics transportation device with image control can run along the required trajectory,locate accurately,and identify the obstacle automatically,to realize automatic management without human.关键词：the image control mold;control;the convey device2442|192|0更新时间：2024-05-07
- 摘要：Line feature is very important to images.In order to extract lines perfectly this paper provides an approach to multiscale image analysis that we call Beamlet analysis.Beamlets have localized scale/location/orientation based on dyadically-organized line segments.Experiments show that the newer method extracts lines better not only in the original images,but also in the noising ones.关键词：Beamlet transform;wavelet transform;line detection3663|203|0更新时间：2024-05-07
- 摘要：A clustering successive projection onto convex sets algorithm is presented for fast point matching.Via feature point clustering,the problem of matching two point sets is converted to that of matching corresponding clusters,which is then solved by a tailored successive projection onto covex sets(SPOCS) algorithm.The resulting algorithm can be viewed as an extention of SPOCS by combining with clustering.Its precision and computational complexity are decided by the clustering radius.Under the condition that the point sets' density is high,by choosing a proper radius,the computational burden can be reduced with only negligible deterioration of precision.Experimental results demonstrate the effectiveness of the algorithm.关键词：clustering;registration;projection onto convex sets3200|217|0更新时间：2024-05-07
- 摘要：Presenting more comprehensive visual information than key-frame,mosaic has been an important element for many vision tasks.This paper proposed a fast and efficient video mosaicing approach based on motion characteristics,which investigated two important issues affecting the quality of mosaic.First,given a video shot,a motion entropy based method is to determine whether the visual content of a shot is suited to be represented by a mosaic.Second,if so,a suitable subset of frames in this shot is selected for efficient and effective mosaicing by constructing a global motion path.Experiments compared to traditional mosaicing have proved that with this approach,not only the visual quality of mosaic is significantly improved,but also the computational time is remarkably reduced.关键词：video mosaic;motion entropy;global motion path;video content representation3499|255|0更新时间：2024-05-07
- 摘要：This paper describes an accurate and efficient algorithm of point cloud auto registration.The algorithm contains two steps: initial registration and precise registration.We use eigenvectors of point clouds to achieve initial registration.For precise registration,an improved ICP(Iterative Closest Point) algorithm based on feature points generated by curvature is introduced.The effect of the algorithm is verified in the applications.关键词：point cloud registration;iterative closest point;reverse engineering7117|332|0更新时间：2024-05-07
- 摘要：As a crucial step in the content-based news video indexing and retrieval system,shot boundary detection attracts much more research interests in recent years.To partition news video into shots,many metrics were constructed to measure the similarity among video frames based on all the available video features.However,too many features will reduce the efficiency of the shot boundary detection.Therefore,it is necessary to perform feature reduction for every decision of the shot boundary.For this purpose,the classification method based on rough sets and fuzzy c-means clustering for feature reduction and rule generation is proposed.According to the particularity of news scenes,shot transition can be divided into three types: cut transition,gradual transition and no transition.The efficacy of the proposed method is extensively tested with news programs over 2 hours and 96.5% recall with 97.9% precision have been achieved.3026|224|0更新时间：2024-05-07
- 关键词：3D model retrieval;2D sketch;view;weighted feature3341|194|0更新时间：2024-05-07
- 摘要：Greyscale image colorizing is a method that transplants color characteristics from a colorized image,source image to a greyscale image,destination image.This method makes the destination image complete with similar color with the source image.We present a fast colorizing algorithm,which introduces the pixel neighborhood relativity into the search procedure for the matched pixels.During the matching,this algorithm searches the current pixel's neighborhood firstly and only searches the whole areas when it failed in the first step.Due to the pixel neighborhood relativity,most pixels can find their matches and only a few of them need to search in the whole area.Therefore it cuts down the cost of searching,improves the speed of colorizing significantly,and in the meantime does not damage the colorizing quality.关键词：colorizing;color space;image processing;neighborhood relativity3610|211|0更新时间：2024-05-07
- 摘要：A 3D model similarity matching algorithm based on spherical harmonics descriptor is proposed in this paper.We first use recursive bisection to partition 3D models,and then we get spherical images by mapping partitions into concentric spheres,last we calculate spherical harmonics descriptors of spherical images to get feature binary trees of 3D models.We can get the similarity among 3D models by matching the similarity among feature binary trees.The experimental results indicate that this approach is robust against rotation transform,noise,model degeneracy etc.,and achieves perfect performance.2784|167|0更新时间：2024-05-07
- 摘要：The development of internet provides a chance for the popularization of GIS.Up to now,many applications of WebGIS depend on GIS software,while it is difficult to deal with and is lack of flexibility because of diversity of GIS software and GIS data.This paper presents another kind of WebGIS using GML organizing geography data and SVG displaying geography data,thus the geography data can be displayed and transmitted on the internet.关键词：WebGIS;geography markup language(GML);scalable vector graphics(SVG)3125|192|0更新时间：2024-05-07
- 摘要：In this paper,Boolean operations on a kind of highly complicated 2D polygons having a large amount of data is concerned.Each polygon is composed of line segments and arc segments,which has a great number of vertices and is of a complicated geometric shape.This algorithm of Boolean operations is introduced specially for these 2D polygons.By area partitioning,the algorithm calculates the intersection points of two polygons.Then it determines the crossing condition at the position of each intersection point,and traces the edges of polygons to form a closed cycle according to the requirement of union,intersection or subtract operation.The results are also 2D polygons.The algorithm is designed from the aspect of practicality,avoiding the complex mathematic reasoning.It can be coded conveniently and improves the efficiency of Boolean operations greatly in practice.关键词：polygons;Boolean operation;LCD design;intersection3638|228|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