最新刊期
卷 13 ， 期 8 ， 2008
- 摘要：Applications for hyperspectral image data are still in their infancy as handling the significant size of the data presents a challenge for the user community. In some cases, especially in satellite data link, limited by the available bandwidth and the onboard storage capacity, lossy compression techniques have to be used. Hyperspectral imagery applications appear different sensitivities to various of degradations caused by different lossy compression methods. The evaluation of quality of compressed hyperspectral images comes in growing interest over the past few years.In the paper we first reviewed some major hyperspectral image compression methods. Second we summarized the existing evaluation approaches including objective distortion criteria, statistical applications algorithms results evaluation and extraction of distortion criteria with same sensitivity. The merits and drawbacks of the approaches were discussed as well.Then a new evaluation scheme based on optimal performance is proposed. In the end, the future trend of the evaluation technique was pointed out, in personal opinion.关键词：compression;hyperspectral image;quality criteria;evaluation5908|426|0更新时间：2024-05-08
- 摘要：In this paper we propose the lifting scheme of true 2D coifman wavele to reduce half computation compared with it directly be used to decomposition and reconstruction 2D data. True 2D coifman wavelet has many favourable properties such as perfect reconstruction, vanishing moments, diamond frequency response and dyadic fractions etc., this makes this method competitive in image compression property preservation. Simply by changing normalizing factor, this lifting scheme can be used for lossy and lossless image compression and at the same time it keeps advantages of 1D wavelet lifting scheme at computation and border extension aspects. The characteristics of prediction and update operators in true 2D coifman wavelet lifting scheme provide guidance for constructing true 2D wavelet filter banks. Simulation results show that true 2D coifman wavelet filter bank has nicer compression property than another 2D wavelet filter bank constructed by the same 1D prototype wavelet filter bank.关键词：true 2D coifman wavelet filter bank;lifting scheme;image coding3412|239|0更新时间：2024-05-08
- 摘要：Context based arithmetic coding is the key component of the new JPEG2000 standard. For significance coding, JPEG2000 designs empirically context model using eight adjacent coefficients and classifies all context events into 9 conditional contexts and every bit plane use the same context classification maps. This paper addresses the improvement of significance coding and context modeling for significance coding in the JPEG2000 which enlarges the context template which was 3×3 square region in JPEG2000.We propose a new scan method based on significant context and exploit optimization techniques for context quantization by minimum description length considering the difference of statistic among sub scans.Finally, we establish different context models for different sub scans. Our experimental results show that the lossless compression performance of our scheme is improved 1.312% upon that of JPEG2000.关键词：JPEG2000 standard;significance coding;context modeling;context quantization3189|298|0更新时间：2024-05-08
- 摘要：The fractal decoding has the merit of simple and fast and so on. But according to the Banach’fixed point theorem, each time the conventional decoding obtains the image that is iterated and transformed, which is not suitable for the situation of advancing transmission and demonstration, such as low bandwidth transmission. And manufacture of computer auxiliary animation, therefore it is necessary to improve it. On account of the fixed parameter, the control of the advancing fixed point theorem is also very limited. In order to be under better control, this article proposes “fixed point theorem of controllable and changeable parameter” to decode based on the characteristic of fractal decoding. The experiment indicates that the method has better and bigger improvement compared to the original method.关键词：Controllable and Changeable Parameter;fractal decoding;fixed point;low bandwidth transmission3631|249|0更新时间：2024-05-08
- 摘要：2D complex image interpolation is decomposed into two real image interpolation procedures based on real part and imaginary part of image. In fact, real image interpolation is a 2D filtering procedure. In this paper we propose a local adaptive interpolated algorithm for 2D real image. Presupposing there is no model, this algorithm utilizes fully local images to estimate autocorrelation function. To minimize the mean square error that interpolated local image, genetic algorithm is introduced in the procedure of optimizing multi variable function. This interpolation algorithm is worth to be applied in ultra wideband synthetic aperture radar (UWB SAR) image interpolation procedures. Two experiments with real data, complex image interpolation and complex image coregistration, have verified feasibility of this interpolation algorithm.关键词：2D image interpolation;autocorrelation function;genetic algorithm;complex image coregistration3358|338|0更新时间：2024-05-08
- 摘要：Compressed video streams are very vulnerable to transmission errors. Channel errors, including bit errors and packet loss, not only corrupt the current frame, but also propagate to subsequent frames. A novel multi directional interpolation scheme based on least squares support vector machines (LSSVM) error correction is proposed to get more satisfying error concealment results. At first, a multi directional interpolation for spatial error concealment is used to restore the missing block. Then LSSVM is trained with the interpolation error distribution of missing block is obtained from large local neighborhood of surrounding pixels of the source image. At last, the more accurate result frame is obtained by adding the estimated error distribution to the restored frame. Experimental results show that the visual quality and performance of the proposed method is improved compared to both nearest pixel linear interpolation and multi directional interpolation.关键词：multi directional interpolation(MDI);error concealment;least squares support vector machine(LSSVM);error correction3697|268|0更新时间：2024-05-08
- 摘要：Classical image scrambling transformation based on high dimension matrix transformation, which is represented by high dimension Arnold transformation or high dimension Fibonacci Q transformation, is capable of preferable security and can change the gray characteristic of the scrambled image, however，there exists some deficiency in practice, such as the long restorable period and not having the full diffuse ability to attacks. To address these problems, in this paper, based on high dimension matrix transformation, an avalanche image scrambling transformation is constructed, which scrambles the original image by inverse transformation and recovers the scrambled image by obverse transformation in order to cut down the iterative times from the scrambled image to the recovered image. Simultaneously, theory and experiments show its strong fragility under attacks and can be used to construct fragile watermarking to ensure digital productions integrality certification.关键词：image scrambling transformation;image encryption;inverse matrix;avalanche effect;fragile watermarking3231|247|0更新时间：2024-05-08
- 摘要：The polarimetric SAR(synthetic aperture radar) image provides a very convenient approach for signal processing and acquisition of information from radar image. Based on statistical formulation of polarimetric SAR image, we present a new approach for speckle reduction using ICA (independent component analysis). In addition, we apply some ICA algorithms to real polarimetric SAR images and compare their performences. The comparison reveals characteristic differences between the studied ICA algorithms, complementing the results obtained earlier. The experimental results show that excellent performence can be achieved, the ENL is high and the image speckle noise is reduced effectively.关键词：independent component analysis;speckle reduction;algorithms comparison3015|272|0更新时间：2024-05-08
- 摘要：A combined model based PDE is proposed here ,which can be used in image denoising. It combines the Total Variation(TV) filter and a fourth order PDE filter by a weighting function τ. TV filter is able to preserve edges and the fourth order PDE filter can overcome the staircase effect in smooth regions. The combined model preserve the advantages of both. Numerical experiments show that the combined model has more improvement on signal noise ratio(SNR) than using only one filter.关键词：total variation;image denoising;staircase effect;PDE3564|268|0更新时间：2024-05-08
- 摘要：Traditional methods for image restoration usually focus on eliminating noises to accomplish image restoration, But for images polluted by strong noises, it is hard to achieve it in this way. Stochastic resonance can use noise to enhance signals. Since the features in a bistable system are induced to the state of aperiodic stochastic resonance resorted to mutual entropy, hence we can restore image with noises use a nonlinear bistable system as a signal processing device. Results in this paper show that images restored based on the method proposed by this paper is more similar to original images than the traditional methods. In addition, the method is robust to variations of the intensity of noises.关键词：stochastic resonance;nonlinear bistable system;image restoration3347|275|0更新时间：2024-05-08
- 摘要：Perceptual image hashing maps an image to a short data string, applicable to image authentication, content based image retrieval, digital watermarking, etc. We propose a new image hashing algorithm using Harris corners and singular value decomposition. Critical feature points robust against gray level modification and image rotation are identified. A prescribed number of large singular values of the image blocks centered at the robust feature points are quantized to compress the data, which represent positions of the points and information of their neighborhood. The compressed data are then coded to generate the hash. The obtained hash is stable to visually insignificant changes due to normal image processing and JPEG coding, while sensitive to excessive changes and malicious tampering. Security of the hash is guaranteed by using secret keys.关键词：image hashing;Harris corner;singular value decomposition (SVD)3042|506|0更新时间：2024-05-08
- 摘要：In the automatic interpretation of visible spectral remote sensing images (RSIs), in order to reduce the time of extracting a specific interested object of man made building, it usually locates coarsely the area of man made buildings first. A remote sensing image has the typical feature that different ground forms have different spectrums: The frequency of building area is higher than that of other areas. Based this kind of difference, an algorithm for locating the building area of RSIs using wavelets transformation and mathematical morphology is proposed. First, wavelets transformation is applied on a RSI to transform different information of ground forms into different frequency bands. Then, the high frequency information is merged, and in the merged image, the area where the amplitude changes rapidly is regarded as a peak, while the area where the amplitude changes smoothly is regarded as a basin. Finally, the building area is extracted based on morphological reconstruction. The experiments show that the proposed algorithm is accurate and fast.关键词：remote sensing images;wavelets transformation;mathematical morphology;morphological reconstruction;region tracking;building area4154|335|0更新时间：2024-05-08
- 摘要：In order to demonstrate the possibility and feasibility that eyebrow can work as a biometric, this paper presents an eyebrow recognition method, the relation of whose accuracy to the observation symbol number and the state number has been preliminarily investigated. The experiment results have shown that the method can achieve an accuracy of 926% in a small scale eyebrow image database taken from 27 persons.关键词：hidden Markov model;vector quantization;eyebrow image;eyebrow recognition3830|242|0更新时间：2024-05-08
- 摘要：In order to describe the object’s shapes steadily by chordal distribution in object’s region, region main chord and it’s perpendicular scanning chords were defined firstly in the paper and their computation method was put forward. Object region chordal distribution is composed of main chord and perpendicular scanning chords. According to main chord and perpendicular scanning chords length and position, the paper defined such effective shape features as chordal area, parallelism degree, edge points symmetry proportion, main chord length and chordal square rate. Finally, the paper used these region chordal features dealing with some simple shapes and cell images.The result showed that these features can distinguish round, ellipse, rectangle, parallelogram and anomalistic polygon. At the same time the cell identifying ratio was improved and misjudgment ratio depressed obviously after the cell region’s chordal features were used in microscopical cell image classifying and recognition system.The system distinguished the shapes among strip tube cell and quadrate rime and round cell and anomalous epithelia effectively.关键词：region main chord;perpendicular scanning chord;chordal distribution;edge points symmetry proportion;region chord parallelism degree3462|225|0更新时间：2024-05-08
- 摘要：In order to study the monitor of cutting tool condition in machining based on computer vision, the experimental system on the state of cutting tool condition monitor system is designed, the image processing technology is introduced into the condition monitoring of cutting tool wear, a new method of judging the cutting tool state is proposed by extracting the cutting surface connected components integer number. The images of cutting surface are collected and preprocessed. The tradition stroke label algorithm is improved, the preprocessed cutting surface images are labeled using the algorithm above, and the connected component integer number is calculated to judge the wear state of cutting tools. Both theoretical analysis and experiments show that there is a strong relativity between the connected component integer number of the cutting surface and the cutting tool wear state, the cutting tool wear condition can be estimated indirectly. And then the objective of monitoring the cutting tool condition is achieved. It is concluded that the method of judging the cutting tool wear condition through extracting the connected component integer of cutting surface is simple, the identification speed is fast, and the cutting tool wear condition can be judged effectively in this way.关键词：image processing;connected components integer;tool wear;study of experiments3413|264|0更新时间：2024-05-08
- 摘要：According to theory of generalized fuzzy sets, the dynamic edge detection algorithm of DICOM Data was given. Many edges were detected by use of the algorithm. That is, not only detect rough edge through ignore detail, but also detect detail edge through enhancement detail. Article ［5］ and ［8］ were specials instance. At last, dynamic edge detection algorithm of RGB image was given.关键词：dynamic edge detection;image enhancement;generalized fuzzy set;linear generalized fuzzy operator;DICOM(digital imaging and communications in medicine)3128|266|0更新时间：2024-05-08
- 摘要：Docking aircraft extraction from the captured image sequence and its shadow detection are the key works in the Visual Docking Guidance System. The algorithm based on Mixture of Gaussians (MOGS) is widely used to subtract static background. However, the real time performance of the MOGS algorithm is reduced remarkably when high resolution image is processed, the “bad tail” phenomenon occurs when slowly moving and large object is extracted, and the shadow of moving object can not be detected. An improved MOGS algorithm based on hierarchical image is proposed, and the problems of bad real time and “bad tail” phenomenon are solved. On this condition, a new shadow detection algorithm based on color character and region character is presented, partially spatial constraints are used, and the shadow of moving object is detected exactly. The experimental results in Visual Docking Guidance System show the validity and the practicality of the algorithms.关键词：visual docking guidance system;mixture of Gaussians;background subtraction;shadow detection3952|256|0更新时间：2024-05-08
- 摘要：Fuzzy C means(FCM) clustering algorithm is one of well known unsupervised clustering techniques, which has been widely used in medical image segmentation. However, when the conventional FCM algorithm is used for image segmentation, no spatial information is taken into account. This causes the FCM algorithm to generate unexpected results of segmentation when dealing with noise of magnetic resonance images(MRI). Considering the intensities of ideal MRI are piecewise constant, we present an improved model to FCM algorithms using multiple objective programming. The algorithm can reasonably use the spatial information of the image and improve the accuracy of segmentation. The new algorithm is applied to both synthetic images and magnetic resonance images and the results show better robustness of our algorithm to noise and other artifacts than the conventional fuzzy image segmentation algorithms.关键词：image segmentation;fuzzy C means algorithm;multiple objective programming;spatial information of image3242|240|0更新时间：2024-05-08
- 摘要：Motion segmentation needs to estimate the model parameters of every motion as well as its supporting region. On the basis of maximum posterior marginal probability (MPM MAP) algorithm this paper presents a new algorithm based on region shrinking to locate the supporting area. In this algorithm pixels of maximum probabilities belonging to a motion are considered to be candidate pixels for supporting region. Then the region shrinking algorithm is used to determine the region of maximum density of the candidate pixels to be the range of supporting area. Moreover, this paper presents a new approach combining the bounding box’s defining with region shrinking to estimate the initial parameters of motions. By motion dividing, motion incorporation and motion elimination the accurate number of motions can be obtained. The results of experiments show the validity of this method.关键词：motion segmentation;MPM MAP;region shrinking3326|242|0更新时间：2024-05-08
- 摘要：Motion estimation (ME) plays an important role in video coding system, since it could significantly affect the output quality of an encoded sequence. This paper proposes a fast adaptive ME algorithm based on the judgment of the predictive motion vector (PMV). This proposed algorithm takes advantage of the correlation of neighboring blocks to adaptively decide the threshold for judging the stationary regions. And this algorithm sorts the PMV into several regions by its direction and uses different search patterns, respectively. Compared with the former algorithm, this algorithm considers the characteristic of different sequences and thus, it is more flexible than before. Experimental result shows that this algorithm reduces computation complexity variously for different sequence under the premise that the PSNR does not lose.关键词：motion estimation;adaptive threshold;judgment of stationary region;directional search pattern3082|217|0更新时间：2024-05-08
- 摘要：A novel pattern classification algorithm called Affine subspace Nearest Points (ASNP) algorithm is presented in this paper. Inspired by the geometrical explanation of Support Vector Machine (SVM) and the nearest point method, in which the optimal separating plane bisects the closest points within two convex hulls, the ASNP algorithm expands the searching areas of the closest points from the convex hulls to their corresponding class affine subspaces. The affine subspaces are taken as the rough estimations of the class sample distributions, and their closest points are found. Then, the hyperplane to separate the affine subspaces with the maximal margins is constructed, which is the perpendicular bisector of the line segment joining the two closest points. The test experiments compared with the Nearest Neighbor (1 NN) classifier and SVM on the ORL face recognition database showed good performance of this algorithm.关键词：affine subspace;support vector machines (SVM);affine subspace nearest point;pattern classification;face recognition3153|246|0更新时间：2024-05-08
- 摘要：This paper presents a new method for unsupervised classification of terrain types using polarimetric synthetic aperture radar data. This unsupervised classification combines the target decomposition theory and the support vector machines. The initial cluster centers are firstly determined by target decomposition advanced by Cloude and Pottier. Then the pixels near to the cluster centers are selected to train the support vector machines using Wishart distribution. The classified results are then used to define training sets for the next iteration if necessary. Finally, by the optimal separating hyperplanes and the kernel method this method obtains extraordinary classification results and neednot much iteration. And the effects of feature vectors consisted of several polarimetric parameters are discussed in detail.关键词：synthetic aperture radar;radar polarimetry;image classification;support vector machine3438|261|0更新时间：2024-05-08
- 摘要：For region matching in stereoscopic images, most algorithms will be affected by the factors such as regions occlusion, regions warping and lighting condition.So the traditional constraint conditions, like order constraint, unique constraint, epipolar constraint and adjacent constraint, may be violated by these cases. In past few years a new algorithm based on the relative position constraint (RPC) between regions is proposed which can overcome most of the problems mentioned above, but it has not satisfying performance in matching occluded objects in the stereo images. Therefore, there remain false matches and miss correspondences in the final results. In this paper, a novel algorithm using both the relative position constraint and the new cost function on the basis of regions center distance and Zernike moments theory is proposed. Furthermore the adjustable weights of cost function are dynamically estimated according to the distance between the centers of two matching regions. Finally, the proposed region matching algorithm is illustrated by three synthesized stereo images, with a comparison to the present algorithm and the superiority of the new region matching algorithm over the present algorithm is experimentally verified.关键词：stereo matching;region matching;Zernike Moments3545|255|0更新时间：2024-05-08
- 摘要：Interpolation of cross sectional medical images is an important procedure in 3D reconstruction．To overcome the shortcomings, which are brought by current algorithms, such as ambiguity of objects’ boundary, high cost of computing. In this paper we propose a novel adaptive image interpolation algorithm that is suitable to medical image processing．This algorithm makes good use of characteristics of anatomical structures and volume relativity and uses the following strategies: firstly to judge whether the interpolation point belongs to ROI, then to get the best point pair according to quality of curve．The experimental results show that the novel method produces new image that is in better accord with the practical demand not only in gray but anatomical structures as well．Moreover, this algorithm is more efficient than current algorithms.Therefore, our algorithm is more suitable to 3D reconstruction．关键词：cross sectional image;relativity;corresponding points;adaptive3568|257|0更新时间：2024-05-08
- 摘要：Generally, normal based shape histogram feature，such as Complex Extend Gaussian Image, is rotation variant for 3D shape. This paper proposed a new kind of normal based shape signature, namely Radius Angle, which can remain invariant under rotation. Then, the Radius Angle Histogram(RAH) is constructed to describe shape contents and used for 3D shape retrievals. The RAH shape descriptor first uses a series of concentric spheres to capture the point distribution information of the given model. Then for points in each concentric sphere, the Radian Normal Angle is computed to extract the local geometry features. Finally, the Radius Normal Histogram is constructed by using the extracted shape signatures. The proposed shape representation remains invariant under rotations. It can be generated from the given 3D model efficiently and easily as well. In addition, this paper discusses the point sampling result’s affect on the final retrieving precision. The voxelization is used to make the sampled point more even over the surface, and better retrieving precision can be achieved by this process. the performance comparisons for the shape benchmark database have proven that the proposed algorithm can achieve better retrieving performance than other similar histogram based shape representations.关键词：normal;polar radius;radius angle histogram;3D shape retrieval3830|224|0更新时间：2024-05-08
- 摘要：3D surface registration is a challenging problem and it has many applications ranging from 3D reconstruction to object detection and recognition. This paper introduces a new 3D surface presentation approach for the 3D surface registration termed angle signature (AS). The proposed method captures the local geometry information of the given point and encodes it into a vector, so it is invariant to rigid transformation. Therefore, it is possible to implement fast registration owing to this simple representation. This method performs robustly against error of estimated normal caused by noise compared with other surface representations. In order to enhance the speed of the algorithm, this approach selects a small set of featured points and calculates their signatures. Registration is based on finding at least three points correspondence between two surfaces and the transformation parameters can be calculated. The results of the experiments show surface registration using angle signature has desirable speed and accuracy.关键词：angle signature;surface registration;point correspondence;object detection3148|292|0更新时间：2024-05-08
- 摘要：This paper proposes a novel system, named Video Booklet, which enables efficient and natural personal video browsing and searching. In the system, firstly representative key frames of the video segment collection are selected through an elaborate booklet generation approach, and then reshaped by a set of pre trained personalized shape templates (such as circle, heart, sector, stamp, etc), consequently printed out on a real booklet or album. When users plan to browse the content of their digital video library, they can firstly browse their booklets in a manner as browsing ordinary photo albums. When they want to watch a certain segment indicated by a key frame in the booklet, they are able to use their camera phones or similar devices to capture the corresponding frame, and send the captured image to a computer via wireless network (such as blue tooth). Thereafter, the target frame is accurately located by a proposed self training and Snakes evolution based active shape model algorithm, and the distortion of the captured image is corrected. Finally the Video Booklet system will automatically find the most similar key frame to the corrected one in the video library and begin to play the corresponding segment for the users. Thereby, Video Booklet builds a seamless bridge between digital videos and analog albums.关键词：video booklet;video retrieval;video summarization;self training;snakes model;active shape models3353|252|0更新时间：2024-05-08
- 摘要：The rapid and precise evaluation of curve & surface parameterizations has wide application. In the paper we present a parameterization technique for cubic B spline subdivision curves with sharp feature. The eigenstructure of the subdivision matrix is analyzed. The subdivision matrix and control points are projected into the eigenspace of matrix. The technique can deal with curve fitting with sharp features directly. The vertices with maximal curvature are as the initial fitting vertices. With the parameterization technique given in this paper, the curve segment with sharp feature is constructed; and the distance between the original data and the curve is also calculated. The fitting vertices are located at the extremities of curve segment; during error estimation it can improve computation speed and avoid mismatching when the target curve is self intersection or when two branches are too close.关键词：sharp feature;parameterization;curve fitting;subdivision3261|238|0更新时间：2024-05-08
- 摘要：Fractal is an effective technology in the modeling of nature scenery, but the strict self similarity of the traditional fractal algorithms has limited their efficiency, Even when a random issue is corporated, the process of genetic growth is still not simulated appropriately. In this paper, the concept of genetic operation are adopted in the traditional fractal tree generating algorithms，the characteristic of the branch is encoded, and the cross over,mutation,selection are operated on the code, The characteristic of the child branch is generated from the genetic operation of the parent branch; they are similar but not identical. As a result, an image that is self similar but not strictly self similar can be generated and the process of genetic growth can be simulated more appropriately. The case study and simulation shows the fractal trees generated by the new algorithm are more stochastic and more real. The genetic operation can effectively remedy the strictly self similar of the fractal algorithm and the limitation of the random issue in the modeling of natural scenery，and this algorithm could be extended to other fractal recursive algorithms.关键词：fractal tree;random perturbation;cross over;mutation;parameter control;random number generator3243|234|0更新时间：2024-05-08
- 摘要：In an augmented reality (AR) system, whether there is a correct real virtual occlusion is a crucial problem, and it will influence whether the compositing of real and virtual worlds is vivid. A new method for real virtual occlusion according to a new design of optical see through HMD is presented. A gridding plane using the OpenGL are drawn to block the virtual object and when the virtual object is overlapped with the real object it will appear that the real object occlude the virtual one. It is easier for us to design programme and is more feasible for a real time system using this method. The experiment results indicate that this method is correct and viable.关键词：augmented;optical see through HMD;real virtual occlusion3856|256|0更新时间：2024-05-08
- 摘要：In this paper we firstly analyze internal structure（refers to structures that are hidden behind outside structures and can not be seen directly） of volumetric data based on the set theory and describe the requirement for the volume rendering method used for rendering internal structures. Based on the requirements, we propose the use of an internal structure region based volume renderer that offers an alternative for visualization of internal structures of volumetric data. The approach can analyze internal structures because it divides the volume data into two subsets: internal structure region (or local region) and context region. The internal structure region provides mechanisms to peel inside the data set for depicting information of internal structures. The user can analyze and understand any region of internal structures through interacting with the internal structure region. The proposed approach can be used to analyze volume data quantitatively and thus improve the analysis quality. The proposed approach shows powerful usefulness in 3D medical image analysis.The experiments show that the proposed approach is effective for analyzing inner structures.关键词：volume rendering;internal structures;medical image analysis3061|401|0更新时间：2024-05-08
- 摘要：Topological structure plays an important role in simplification, transformation, surface reconstruction, compression and transmission, matching and retrieval, shape analysis and understanding, and applications of point cloud data. It is one class of the key characters of 3D point cloud data. Relevant concepts, methods and techniques of topologically structural representation and computation are surveyed and summarized in this paper. Typical techniques and their latest developments are remarked. Practical applications of topologically structural representation and computation of point cloud data and our some latest results are presented. Our perspectives of these problems and techniques are shown.4299|275|0更新时间：2024-05-08
- 摘要：Global translation preferred active contour is much more suitable for object tracking. The prior information of active contour with global translation preference can be interpreted as the velocity field along the evolving contour with equivalency preference. According to this, a simple gradient flow which preferes global translation is acquired by defining a new inner product on the set of perturbations of a curve. The new inner product is obtained by adding the variance of the perturbation of a curve to the inner product relative to the 〖WTHX〗H〖WTBZ〗0 active contour. The active contour relative to the new inner product is called variance active contour. In contrast to 〖WTHX〗H〖WTBZ〗1 active contour generated by convolution of 〖WTHX〗H〖WTBZ〗0 active contour and certain kernel functional, variance active contour needs no convolution and is a weighting sum of 〖WTHX〗H〖WTBZ〗0 active contour and corresponding average gradient flow. Thus, variance active contour can be implemented much faster and easier than 〖WTHX〗H〖WTBZ〗1 active contour. We also compared 〖WTHX〗H〖WTBZ〗0, 〖WTHX〗H〖WTBZ〗1 and variance active contours in space and frequency domains.关键词：variance active contour;Sobolev active contours;inner products;gradient flows;G〖AKa^〗teaux derivative3762|233|0更新时间：2024-05-08
- 摘要：Scene recognition is a key problem in mobile robot topological navigation. For unknown environments, a natural scene recognition approach based on visual local salient regions is presented. Firstly, a feedback saliency detection model (FSDM) is presented to carry out bottom up scene image analysis. Then, according to the salient positions, automatic scale selection is realized based on fractal dimension to build the local salient regions with appropriate size. Those salient regions are represented by 3 invariant features of gradient orientation, moment and canonical hue. They are used for scene recognition in terms of their match ratio. Experiments show that FSDM can obtain higher accuracy. The scene recognition experiments in both indoor and outdoor environments show that the approach has high stability and tolerance compared to the method based on global appearance when scale and viewpoint etc changed. The accuracy of recognition for static scene is higher.关键词：scene recognition;local salient regions;FSDM (feedback saliency detechion model);scale selection;fractal dimension3820|284|0更新时间：2024-05-08
- 摘要：According to vehicle characteristics can not be extracted at night, a vehicle classification method based on improved fuzzy constraint satisfaction approach is proposed. Through analysis and pretreatment vehicle’s lumps from monitor video at night, the membership function’s parameters of fuzzy constraint algorithm were adjusted; thereby vehicle’s types were recognized from the constructed motion trajectory image. The experimental result of the system shows that the accuracy of vehicle classfication was improved at night, when we involve vehicle lamp information.关键词：fuzzy constraints satisfaction;vehicle recognition at night3916|252|0更新时间：2024-05-08
-
An Intelligent Surveillance System for Freshwater Sea route Based on Multi vision Information Fusion
摘要：In this paper we discussed about monitoring the traffic of freshwater sea route by mean of dynamic image understanding technology, in which multi vision sensor data amalgamation technology is used to capture the concerned areas of tracked ships. With real time collecting, processing, tracking, and analyzing the video flow, some important traffic information including speed, number of ships, and the density of the ship in route are automatically obtained. The captured ship images are analyzed to determine license plates and load of the tracked ships and to obtain some basic data useful for river management department’s decision making for management and planning. In addition, the surveillance area and object can be defined by users so that the system processing time is reduced while the ability to real time processing, reliability, and robustness are improved. Experimental results show that the system can realize tracking the ships stably, can capture and recognise the specified part of the ship.The accuracy rate of recognition is high, at last the system realizes calculating the tonnage.关键词：video intelligence surveillance;ship tracking;dynamic image understanding;video data amalgamation;statistic of the number of ships3618|244|0更新时间：2024-05-08 - 摘要：Phylogenetic study needs to learn both external and internal morphology of specimens which requires inevitable destructive dissection. The rare and important specimens nonetheless do not allow any destructive operation. The present work used CT and 3D visualizing technique to image a fossil giant panda skull, the earliest and the most complete one so far in the world, as well as an extant giant panda skull for comparison, segmented and reconstructed the 3D images and virtually dissected their paranasal sinuses for new anatomic data.关键词：3D virtual reconstruction;noninvasive dissection;giant panda;paranasal sinuses2991|246|0更新时间：2024-05-08
- 摘要：Diffraction enhanced imaging (DEI) is a powerful phase sensitive technique that generates improved contrast of soft tissues compared to conventional radiography, and it can allow the visualization of nearly scatter free images displaying both absorption and refraction properties. Multiple image radiography (MIR) is a significant improvement over DEI based on a statistical analysis. MIR is more robust to noise and produces an additional image that depicts ultra small angle X ray scattering (USAXS), and the USAXS image conveys the presence of microstructure in the object. Therefore, MIR simultaneously produces three images, depicting separately the effects of absorption, refraction and ultra small angle scattering of X rays. Furthermore, the absorption, USAXS and refraction images can be fused together and several kinds of information can be integrated in one image. In this work, three MIR images of biological samples are obtained and fused together, and the results show that MIR images can clearly display the inner details and structures of biological samples.关键词：phase information extraction;diffraction enhanced imaging;medical imaging;multiple image radiography;image fusion3255|243|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