最新刊期
卷 13 ， 期 2 ， 2008
- 摘要：In previous content-based image retrieval algorithms, the most prevalent and convenient method in representing images is to extract low-level content features such as color, texture, shape or spatial information. But using only one low-level feature independently ignores the relevancy and coherence between features will cause a limitation on making the most of information contained in an image. The usage of single feature also confines the ability of multiple features to cooperatively illustrate images. Fusion of two or more low-level features will make a connection between features and enhance the efficiency and accuracy of image representation. Feature fusion is a trend of research in content-based image retrieval. In this paper, an up-to-date overview of low-level feature fusion algorithms is presented. In addition, a classification system of fusion algorithms is established based on the fusion levels and the content of fusion. The existing problems and open questions in this field are also indicated.关键词：content-based image retrieval;low-level;fusion;classification system3516|397|0更新时间：2024-05-08
- 摘要：Iterative decomposition and reconstruction are needed in Mallat algorithm. In order to realize perfect reconstruction, finite-length signals must be extended to some extent before they can be transformed. The algorithm based on periodic boundary-extension always can be seen in the literature. Symmetric boundary-extension has better performance than periodic method in image processing, whereas the matrix transform method based on symmetric boundary-extension is seldom mentioned in the literature. A method of constructing decomposition and reconstruction matrices with arbitrary wavelet transform depth in mirror-symmetric boundary-extension is proposed for wavelet transform in matrix-vector multiplication, and the condition for perfect reconstruction of Mallat algorithm is proved. As an example, the base vectors and base graphs of Bior33 wavelet were given. The application of wavelet transform matrices in the wavelet-based image processing can avoid iterative operation, simplify the calculation and meanwhile reduce the edge effect evidently.关键词：wavelet transform;mirror-symmetric extension;biorthogonal wavelet;Mallat algorithm3935|410|0更新时间：2024-05-08
- 摘要：A fragile watermark is designed using integral wavelet transform combined with hash function in order to verify the integrity of medical images. Even 1 bit of change in the picture can be sensed. Furthermore the region distorted can be easily oriented in this algorithm without an original image. Compared with the traditional digital watermark based on wavelet transform, the integer wavelets not only simplify the calculation but also improve the quality of watermarked image. And the hash function constructed with MD5 algorithm increases the security and ability of tamper localization of watermark. The results of experiment indicates that watermark scheme is highly sensitive to the distortion of the image. Since it requires a key during both the insertion and the extraction procedures, it has high security and the whole process is simple and accurate.关键词：medical image authentication;fragile watermark;integral wavelet transform;quadtrees;Hash function3261|272|0更新时间：2024-05-08
- 摘要：The Ultrasound images are usually degraded by speckle which brings much difficulties in diagnosis and treatment. In noisy images, the Gaussian convolution is often applied to smooth image and then gray differential is calculated to detect image edges. A structure tensor based on Gaussian convolution is devised and introduced to anisotropic diffusion in order to reduce speckle. Experiment shows that the novel anisotropic diffusion based on structure tensor can reducing speckle, detect and preserve edges and useful details effectively.关键词：speckle;anisotropic diffusion(ANDI);structure tensor3260|260|0更新时间：2024-05-08
- 摘要：A new adaptive contourlet transform-based technique for speckle removal from SAR images is presented. The distribution of the SAR image’s contourlet coefficients is mainly decided by speckle noise and useful signals corrupted. The analytic model for these distributions is proposed. The shrinkage factor for de-specking is estimated with the neighboring reference contourlet coefficient in every sub bands. Finally, the comparison of performance of Lee filter, Froster filter, Gamma filter, wavelet-based de-speckling and contourlet transform-based de-speckling is provided for both simulated and actual SAR images. It shows that the contourlet methods strongly suppress speckle, while preserving image details and sharpness.关键词：Contourlet transform;SAR image;speckle;adaptive shrinkage estimator3390|267|0更新时间：2024-05-08
- 摘要：BayesShrink is one of the best algorithms for wavelet thresholding denoising, while WienerChop improves VisuShrink by Wiener filtering in wavelet domain. We studied the denoising method uniting BayesShrink and WienerChop. The combined algorithm has smaller mean squared erroe(MSE) and higher signal to noise ratio(SNR) than BayesShrink or WienerChop. It integrates the advantages of the two algorithms, and improves the problems which images are smoothed overly by WienerChop and BayesShrink retains some noise artifacts. It can visually obtain more pleasing denoised images.关键词：adaptive method;image denoising;image restoration;wavelet thresholding3159|353|0更新时间：2024-05-08
- 摘要：Due to their complexity, traditional edge-based algorithms of image interpolation are hardly used for real-time image processing. And non-edge based interpolation algorithms usually lead to the target images blurring because of their low-pass effect, so the visual effect of the interpolated images is unpleasant. In this paper, a novel adaptive edge-based algorithm of image interpolation is proposed to eliminate the defect. According to the second-order difference of the consecutive pixels, the source pixel group is automatically selected to compute the corresponding target pixel gray value. Several algorithms of image scaling are implemented in Matlab. Then the visual appearance and the edges of target images are compared, and the complexity of the algorithms is analyzed. The implementation circuit architecture for our proposed algorithm is devised and the algorithm is realized by Verilog HDL. Finally, the synthesized code is verified by downloading to FPGA on the evaluated board. The results show that our proposed algorithm achieves vivid edges of target images, and has lower computation complexity and easy for hardware implementation. So the algorithm can effectively interpolate images and adapts to real-time image resizing.关键词：real-time image processing;adaptive algorithm;image interpolation;image edge3994|306|0更新时间：2024-05-08
- 摘要：CAVLC decoding methods known by people include bintree, full code table, and Hashemian decoding methods, etc. all of which lay emphasis on only one aspect of decoding: decoding speed or memory space, and thus the general capability can not be improved effectively. In view of that, the paper puts forward a new and fast decoding method which by means of the automatic distribution of the code table and the transfer of the code table address, improves the decoding speed in the limited memory space. As the experiment shows, in the same-size memory space, the speed of this new method is 15 times as fast as that of the tradition ways so that the new one is more suitable to H.264.关键词：CAVLC;bintree;full code table;Hashemian method;the automatic distribution of the code table;the transfer of the code table address3352|243|0更新时间：2024-05-08
- 摘要：Problems as low speed or inaccurate results in the process of line detection with Hough Transform remain unsatisfactorily solved. This paper puts forward a new look-up table to decrease the computation distinctly and keeping the highest precision. It offers the possibility in real-time applications especially in large image.关键词：Hough transform;parameter space;precision;table look-up4670|300|0更新时间：2024-05-08
- 摘要：The key of high temperature measurement by using CCD image sensor is to recognize the measured object accurately. From analyzing the characteristic of the noises in high temperature radiator color image, an image recognition method of high temperature radiator is proposed, which is based on the color information of a high temperature radiator. By segmenting the red and green color images, the various noises in high temperature radiator color image are reduced or even eliminated. And then, by synthesizing the results of the segmented red and green color images, the high temperature radiator is recognized accurately. Finally, in order to smooth the image’s edge, the morphology method in mathematics is applied to process the segmented image and remove the dissociations and narrow holes. The experimental results show that the method can recognize high temperature radiator with high accuracy and practicability.关键词：object image recognition;CCD image sensor;temperature measurement3360|275|0更新时间：2024-05-08
- 摘要：In this paper a method which can improve the performance of face recognition system using color Gabor features is presented. First, quaternion is used to describe the color information, considering that Gabor filters have desirable characteristics of spatial locality and orientation selectivity, and they are extended to quaternion space. Then utilizing the convolution of the key points and the Gabor filters to extract features, by doing this the gray Gabor features are extended to the color ones. In the end, for the extracted features, we used PCA for dimension reduction and SVM for recognition. The experiment carried on Color FERET Database and the result utilizing ROC curve for cross-validation show that the use of color texture information can improve the efficiency of face recognition system markedly.关键词：color Gabor;color texture;feature extraction;SVM3851|293|0更新时间：2024-05-08
- 摘要：Polarization axis should match when two Panda polarization maintaining fibers connect, in order to getting high accuracy identification of the polarization axis, an accurate computer image identification algorithm was presented to identify the polarization axis of the Panda polarization maintaining fiber. To validate the algorithm, program was developed to perform image analysis and processing with LabVIEW software. The practical data of application indicates that they are very useful to improve the identification precision of polarization maintaining fiber and raise productivity of fiber connector with the technologies of computer-aided vision detection and modern measurement & control.关键词：Panda polarization maintaining fiber;computer image identification;LabVIEW;fiber jumper connector;edge detection algorithm3585|271|0更新时间：2024-05-08
- 摘要：To solve the problem that it is difficult to directly detect the object in the water due to flares and cloud shadows, this paper proposed a water wake recognition method based on Multi-Channel Gabor filters, and BP neural network. First, we select sample sub-images of same sizes with wake texture and without wake texture, then, we obtain feature images using a group of Gabor filters and calculate the mean and variance of feature images to acquive, the input vectors and train the BP network. Secondly we divide the whole image into sub-images with the same size as the first step, calculate mean and variance of Gabor feature images, caculate the input vector and judge whether the sub-image contain a wake texture by the trained BP network in the first step. We obtain a binary image by the classify results of the whole image, detect lines using Hough transform and judge whether there is a wake in the whole image. From experiment results, it is proved that the proposed algorithm can attain the wake texture precisely.关键词：Gabor filters;BP neural network;Hough transform;texture fetching3644|301|0更新时间：2024-05-08
- 摘要：It is very important to extract water and settlement areas from SAR images in practical applications. In this paper the single band and single-polarization Radarsat-1 SAR image is used to water and settlement area extraction. Firstly, the statistic structure information of sample image is analyzed using semi-variogram to determine the optimum parameters for textural information extraction. In order to establish the multi-dimension feature space for image classification, the textural measures such as mean, angle second moment and entropy have been calculated based on grey level co-occurrence matrix method. Then the water and settlement area information can be enhanced effectively using false color composite method. Three types of sample such as water, settlement and other are collected as training samples, and the image data are processed using support vector machine classification method. Finally, the image fusion on the target level between classification result image and NDVI image is conducted in order to eliminate the mountain influence, and the water and settlement areas are extracted accurately with a total classification accuracy of 8257%, and Kappa coefficient of 0.58.关键词：synthetic aperture radar(SAR);semi-variogram;grey level co-occurrence matrix;support vector machine(SVM)3681|292|0更新时间：2024-05-08
- 摘要：There are sometimes shadows, such as cars、trees and buildings ,on road surface images, which make it difficult to abstract and recognize the features. After analyzing the shortages of Center/Surround Retinex(CSR) algorithm and shadow removal approaches based on Retinex in existence, an Anisotropic Diffusion Center/Surround Retinex(ADCSR) is presented to solve the problem. First anisotropic diffusion based on PDE is introduced to ADCSR, further a new anisotropic diffusion scheme based on “Edge Degree”(ED) is presented, which avoids the embarrassment to select different parameters such as gradient threshold. Theoretic analysis and experimental results show that the effectiveness of the approach.关键词：road surface image;shadow removal;anisotropic diffusion;Retinex;PDE3072|294|0更新时间：2024-05-08
- 摘要：In this paper， we propose a new method for 3D discrete curve smoothing by adopting an intrinsic representation for discrete curves. With this representation， a piecewise 3D curve is defined by lengths of edges， angles between edges and the positive x-axis， and angles between edges and the positive z-axis. For a noisy curve， we first filter the two angle sequences of the curve by bilateral filtering method. Then， we obtain the smoothed vertexes of the curve by solving an objective function under the constraint of the two angle sequences. By this algorithm， not only can main features of the original curve be preserved well， but also the smoothed curve no longer suffers from shrinkage.关键词：smoothing;bilateral filtering;intrinsic representation;curve shrinkage3165|273|0更新时间：2024-05-08
- 摘要：In this paper, a novel Harris corner detection algorithm based on condition number is proposed. Firstly feature points are extracted using Harris corner detector which usually leads to either missing significant corners or detecting false corners due to noise. To overcome these drawbacks, condition number, which investigates the relationship between noise and the transformation of images, is used to screen out the steady points. Then a new match method called Random Sample Consensus(RANSAC) is applied to determine matching pairs. A number of experiments demonstrate the accuracy, efficiency and robust of the approach.关键词：condition number;Harris corner detection;RANSAC;image registration2919|276|0更新时间：2024-05-08
- 摘要：Being an efficient method of information fusion, multisensor image fusion has been used in many fields such as machine vision, medical diagnosis, military applications and remote sensing. In this paper, PCNN is introduced in this research field for its interesting properties in image processing, including segmentation, target recognition et al., and a multisensor image fusion scheme based on modified PCNN is proposed. The basic idea of the scheme is to segment all different input images by PCNN and to use this segmentation to guide the fusion process. At the same time, a new region feature, which emphasized the salience of target regions and its neighbors is proposed. Focusing on the famous difficult problem of PCNN, how to determine PCNN parameters adaptively, an adaptive PCNN parameters determination algorithm is also presented in this paper. Experimental results demonstrate that the proposed fusion scheme outperforms the multiscale decomposition based fusion approaches, both in visual effect and objective evaluation criteria. It avoids some of the well-known problems in pixel-level fusion such as blurring effects and high sensitivity to noise, particularly when there is mis-registration of the source images. The research fruits have certain value on the theory research and practical application of PCNN.关键词：multisensor image fusion;pulse-coupled neural network(PCNN);parameter determination;objective evaluation criteria3456|277|0更新时间：2024-05-08
- 摘要：Feature matching is a fundamental problem in computer vision. There are geometrical deformations, named plane homography, between corresponding feature windows on different images of the same patch on scene surface. But most of the state-of-the-art matching schemes approximate the homography with an affine transformation model, namely affine invariant features. The linear character of affine model not only reduces the complication of matching process, but ensures a good and stable convergence of the unmanageable iteration. However, there are no quantificational discussions about this approximation by far. In this paper, an overview of invariant features under different geometric groups was given first, and then a quantificational analysis of affine invariant features was proposed. The analytic expression of location error was deduced by our novel method, named normalization of conic equations. Experimental results of real images demonstrate the correctness and necessity of our method. Finally, some valuable conclusions and suggestions based on the above analysis are deduced both on theory and in practice.关键词：feature point matching;homography;affine invariant;elliptical conic;location error3390|332|0更新时间：2024-05-08
- 摘要：To register point sets automatically, a method was proposed in this paper. It used the relational matching to get the counterpoints of marked points based on the topology and introduced the Rodrigues parameters to describe the transition matrix. The matrix was solved simply and steadily by using a temporary variant. Finally a registration experiment was presented.关键词：automatic registration;marked point;relational matching;Rodrigues parameters3616|249|0更新时间：2024-05-08
- 摘要：To improve the efficiency of 3D model retrieval, an algorithm for 3D model retrieval based on integer medial axis skeleton was proposed in this paper. The integer medial axis skeleton and the geometric information of skeleton point were obtained after the preprocessing of the model. The binary tree of this skeleton was acquired by decomposing the skeleton into a set of blocks by spatial region. To describe the influence of different node of the skeleton binary tree to the similarity matching, the feature weight was defined for each node. Furthermore, the weights were determined by corresponding skeleton region of the 3D model. Finally, a coarse-to-fine strategy was presented to calculate similarity between different 3D models. Differing from other algorithms applied in 3D model retrieval, this algorithm extracts statistical features as well as topological features. The experiments have been carried on a standard testing database of 3D models, and the results show that this algorithm can achieve better retrieving efficiency than other algorithms.关键词：3D model retrieval;feature transform;integer medial axis skeletons;skeletal binary tree3210|279|0更新时间：2024-05-08
- 摘要：Shape is a very important visual and semantic feature used to depict image, and it can be revealed by image pixels’ regional distribution. This paper proposes a region-based shape representation, a new “density distribution feature (DDF)”. After shape center orientation and region partition, two M dimensional feature vectors are got. The first feature vector represents the relatively density of object pixels within each sub-image. And the second represents the difference of relatively density in the direction of radial coordinates. When matching the similarity, we first used the Gaussian model to normalize the two dimensional feature vectors. Then we integrated them to calculate similarity distance. The experiments results showed that this shape feature can depict the image well and is invariant to translation, scale and rotation. The paper also evaluated the effectiveness of the proposed descriptor with respect to Moment Invariants.关键词：shape representation;density distribution feature;invariant performance;binary image retrieval3607|320|0更新时间：2024-05-08
- 摘要：This article researches the clipping plane technology during volume visualization in underwater acoustic energy fields to study the influence to naval battles, and brings forward two methods to create clipping plane according to different requests of tactics. These two methods supply each other, and commanders can get the underwater sound field environment and battle information quickly.关键词：underwater acoustic energy fields;volume visualization;clipping plane;volume rendering3498|292|0更新时间：2024-05-08
- 摘要：A novel technique is presented for rendering large-scale volume datasets interactively on general purpose PC hardware. To circumvent the limited texture memory for texture based volume rendering, the dataset is partitioned into the bricks with reasonable size. The bricks are loaded to the graphics hardware dynamically and rendered using 3D texture mapping. During the rendering only one brick resides on the texture memory. Additionally, the sophisticated PC graphics hardware functionality is utilized to estimate the gradient on the fly avoiding the huge memory consumption in previous approaches. Using a prototype implementation of the algorithm, we are able to perform fast data loading and interactive visualization for the large datasets on a single standard PC.关键词：volume rendering;3D texture mapping;texture partitioning;graphics process unit3194|271|0更新时间：2024-05-08
- 摘要：In order to efficiently reconstruct 3D medical image and clearly display the detailed information of inner structures and the inner hidden interfaces between different mediums, an improved volume rendering optical model(IVROM) for translucent volume rendering is proposed, which can be easily realized on a general PC. On the basis of classic absorption and emission model, the effects of volumetric shadows, direct and indirect scattering are also considered in the proposed model Moreover, the implementation of the translucent volume rendering method incorporating the IVROM model and Shear-Warp is described detail. Experiments demonstrate the good performance of the proposed method.关键词：optical model;Shear-Warp;medical image;translucent volume rendering;shading3291|280|0更新时间：2024-05-08
- 摘要：A novel illumination-invariant change detection algorithm based on Cauchy distribution is proposed. The intensity ratios between two images are used as the feature to model and subtract background. The distribution of the intensity ratios between corresponding pixels of two background images follows Cauchy distribution, assuming that some observed temporal intensity variation of each pixel in background images are caused by white noise. The intensity, hue and saturation in the YCbCr color space are employed to recognize and eliminate shadows in video sequences. Finally, experimental results demonstrate that the proposed algorithm can tolerate the whole or local quick or slow changes in illuminations, and can filter noise caused by small motion in scene background.关键词：change detection;moving object detection;background modeling;shadows;Cauchy distribution;surveillance3191|276|0更新时间：2024-05-08
- 摘要：Fast and robust image registration is an important research problem in motion video processing. In this paper, we present a novel invariant feature based automatic image registration method to deal with the large image transformation, rotation, scale and illumination changes. The algorithm includes three parts: invariant feature detection, description and matching. First, the multi-scale space of the image is created via a multi-level box filter, and then the feature points are detected in scale space by considering the quality and special distribution simultaneously. Second, we use the Principle Component Analysis to descend dimension of SIFT(Scale invariant feature transform) for feature description. Finally, based on the principle component of the descriptors, a cascade filtering is designed to speed up the feature matching. Experiments with motion videos captured by the dome camera and the Unmanned Aerial Vehicle demonstrate that the proposed algorithm has satisfied performance, and it can provide a solid foundation for further processing such as moving object detection, tracking and classification from moving platform.关键词：image registration;invariant feature;multi-scale space;PCA;cascade filtering3525|319|0更新时间：2024-05-08
- 摘要：This paper presents an efficient background subtraction algorithm using multiple scene models to cope with variations of noises in a background. A mechanism has been developed to add and delete scene models so that the distribution of the models is adaptive to the background characteristics. The calculation for the model parameters has been optimized so as to avoid time-consuming floating point calculation. We introduced the living time and recurrent frequency to the models so that the algorithm can suppress high frequency background noises effectively by controlling the model recurrent frequency. Experiments using video data have been conducted to compare the performance of our algorithm with that of the mixture Gaussian model algorithm. The experimental results demonstrated that our algorithm can extract the foreground contour more precisely, efficiently and with less memory, while maintaining the advantages of the mixture Gaussian model algorithm. It was also found that high frequency noises that cannot be rejected by the mixture Gaussian model can be suppressed.关键词：VSAM(video surveillance and monitoring);background subtraction;mixture Gaussian model;fast algorithm3481|251|0更新时间：2024-05-08
- 摘要：This paper proposed an adaptive background model for motion detection based on statistic information of probabilities. The approach can select thresholds of foreground and background adaptively and adapt to different scenes without training samples and human concerned. By using statistic information of historical pixels to update background can deal with moved/inserted objects in background. Meanwhile, a background model updated according to statistical characteristic is also provided in this paper. It gets ideal background and good detection results. Experimental results demonstrate the proposed algorithms can get relative good background and improve detection results for different scenes.关键词：motion detection;adaptive background;updated by probability;statistic information3069|282|0更新时间：2024-05-08
- 摘要：In this paper a novel Multi-layer Homography algorithm for background modeling under free moving camera environment is proposed. Background is composed of many planes. RANSAC is used to find these different planes, called Multi-Layer Homography. Each pixel definitely belongs to certain plane. Transformed by the corresponding Homography, each pixel in each frame can find its match in the subsequent frame if it occurs in the shared view of these two frames. Thus, background model can be built. Experiment shows it is effective for background modeling under free moving camera environments.关键词：Homography;background modeling;correspondences3255|269|0更新时间：2024-05-08
- 摘要：Moving target detection is one of the critical issues in video surveillance. Background subtraction is a typical approach to detect moving targets in surveillance using only one static camera. The performance of background subtraction algorithms depends on how to construct background model. In this paper, codebook is used to estimate the background model using a long clip under limited memory. First, an initialization codebook is constructed for each pixel. Second, codewords in the initialization codebook are selected to construct background codebook based on the assumptions about the background model. The image can be partitioned to foreground and background according to the background codebook, and in the meanwhile codebook is updated correspondingly. Experimental results demonstrate the effectiveness of the method presented in this paper, even when there are moving clutters in the background and illumination variations.关键词：moving targets detection;background subtraction;codebook3538|285|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