最新刊期
卷 18 ， 期 8 ， 2013
- 摘要：With the development of software and hardware technique, computer vision has become a hot research fields in image processing. Scale invariant feature transform (SIFT) is one of the most successful vision algorithm nowadays and it is widely studied by the computer vision community because of its unique features.SIFT is scale invariant, rotation invariant and illumination invariant. However, it also has some problems such as it is only part affine has a rather the high computation complexity. Many extended or modified algorithms of the SIFT are developed unceasingly. In this paper, we summarize the history, the evolved processing, and the application of the SIFT and compares those algorithm effects. At last, the paper discusses the feature direction and provides reference for computer vision researchers.关键词：scale invariant;scale invariant feature transform (SIFT);computer vision;image match4121|348|0更新时间：2024-05-07
- 摘要：Aiming at the problem of H.264 inter frame mode selection algorithm, a fast inter frame mode selection algorithm is proposed in this paper. The algorithm selects the macroblock mode by using homogenous analysis of an 8×8 block motion vector, to narrow the scope of the candidate modes and decrease the computational complexity of the H.264 inter frame mode selection algorithm. Meanwhile, the algorithm uses mode correlation and motion vector merging to optimize the UMHexagonS algorithm, which could reduce a large number of motion estimation search points, and obtain higher coding efficiency. The experimental results show that compared with inter frame mode selection algorithm using UMHexagonS, the proposed algorithm can reduce about 50% motion estimation time with only negligible change of PSNR and output stream bit rate.关键词：H.264/AVC;mode selection;UMHexagonS;mode correlation;motion homogeneity2774|254|0更新时间：2024-05-07
- 摘要：A new algorithm for vehicle velocity calculation though automatic and dynamic camera calibration is presented in this paper.The algorithm is based on distance information in the light transmission model and a camera linear model under homogenous fog.Unlike previously published works,the factor of homogenous fog is added into our traffic model. Only road plane and moving foreground are included,while painted lines and other traffic prior information could be neglected. Three major steps construct our algorithm. First, an area search method (ASM) based on activity map recognizes the current weather condition.The current weather condition is assumed to be homogenous fog if the average pixel value from top to bo-ttom in a selected area of interest changes in the form of an edge spread function. Second,using a dark channel prior algorithm a transmission image is calculated.Intrinsic and extrinsic parameters of the camera are calculated based on the para-meter calculation formula especially for our monocular model. In this step, eight key points with special transmittance for generating necessary calculation equations are selected to calibrate the camera. The mean velocity is retrieved based on velocity calculation formula by transforming coordinates from the image plane to the world coordinate plane. At the end of this paper, calibration results and vehicles velocity data for nine vehicles in different weather conditions are given. Comparison with other algorithm verifies the effectiveness of this proposed algorithm.关键词：vehicle velocity estimation;dark channel prior;camera calibration;homogenous fog2442|318|0更新时间：2024-05-07
- 摘要：Nowadays, advances in videophone place higher quality on human facial images. To tackle this issue, a new spatial error concealment algorithm based on the facial features (FFSEC)is proposed in this paper. The algorithm is constituted by three steps. First, the lost blocks are classified by the edge detection technique so as to finish the preliminary error concealment. Second, the facial area is extracted by the facial partition technique. Finally, an interpolation is implemented according to the relative position and specific feature of the five organs, where the position of eyes is obtained by the pattern recognition technique at first. JM11 model of H.264 standard is used to evaluate this FFSEC algorithm. Simulation results showed that compared with the existing adaptive spatial error concealment based on face features (BFASEC)algorithm, when the packet dropping probability is too large, the FFSEC algorithm achieved better visual quality, with 0.5~1.5 dB gain in PSNR.关键词：error concealment;spatial;H.264;facial features2751|246|0更新时间：2024-05-07
- 摘要：Most existing forgery detection algorithms have the weak robustness to post-processing operations of the copy-move forgery regions. For the detection of image regions forged by copy-move, we propose a detection algorithm based on radial harmonic Fourier moments in this paper. First, the detected image is divided into multiple overlapping blocks with the same size. Second, radial harmonic Fourier moments of each block are extracted as a feature vector and all extracted vectors are sorted. Last, the similar blocks are determined by a pre-established threshold, the error similar blocks are eli-minated by using a threshold of the displacement vector, and the tampered region is located. Experimental results show that the proposed algorithm deals well with various post-processing operations of image region copy-move forgery, including noise addition, blurring, rotation, translation and so on. In addition, our algorithm has a better detection rate than that of HU moment-based method.关键词：image forensics;region copy-move tampering;forgery detection;radial harmonic Fourier moments3119|228|0更新时间：2024-05-07
- 摘要：To fast and accurately detect videos that were tampered, the human visual model is introduced to our algorithm. A multi-level and multi-granularity algorithm to detect and locate video tampering is presented in this paper. The random block sampling technology is used, and video structure perceptual features and time-domain perceptual features of video images are extracted. Then, the unidirectional abstract of the hash theory is used to quantify perceptual features, and the video abstract hash value is obtained. The similarity matrix is applied to give a multi-level and multi-granularity detection and location for tampered data. Experimental results show that similarity fitting diagram can reflect the attack power and the attack site of video tampering. The proposed algorithm shows better precision and positioning accuracy.关键词：video tampering;video perceptual hashing;multi-granularity location;perceptual features3629|290|0更新时间：2024-05-07
- 摘要：A key frame extraction algorithm based on the visual attention model is proposed for lane surveillance video. First, the top-down method is used to detect moving objects whose position saliency is decided by the clearest position of license plates and vehicles. Then, within the moving objects, the bottom-up method is used to calculate the moving orientation and moving intensity saliency of these moving objects. Next, the visual attention curve is fused by a simple adaptive linear mode. Last a derivative curve is generated, from which the frame with the most salient value in those zero-crossing points from the positive to the negative on the derivative curve is selected as key frame. Experiments show that the key frames extracted by the proposed algorithm not only include the optimal or suboptimal positions of all passed vehicles, but also include on-street parking, speeding, reverse driving, and other traffic incidents. The results are consistent with the traffic observers’ visual perception and conducive to the extraction of vehicle static features to form the traffic video features database.关键词：key frame;visual attention model;lane surveillance;attention curve3572|364|0更新时间：2024-05-07
- 摘要：Recent research has shown that a better recognition performance can be attained through representing image sets as points on Grassmannian manifolds. However, the conventional discriminant analysis methods based on such manifolds take into account only the statistical information of labeled samples and suffer from ignoring unlabeled samples. To address this issue,a new method based on manifold regularization, called semi-supervised discriminant Analysis on Grassmannian Manifold(SDAGM), is presented and applied to the image sets recognition problem. In SDAGM, a nearest neighbor graph is constructed to capture the local geometrical structure of all samples on the Grassmannian manifold and incorporates them into the objective function of discriminant analysis on Grassmannian manifold as a regularization term. Not only does the proposed algorithm consider the label information, but it also uses a consistency assumption. The feasibility and effectiveness of SDAGM are verified on several standard data sets with promising results.关键词：Grassmann manifold;image sets;discriminant analysis;semi-supervervised2880|248|0更新时间：2024-05-07
- 摘要：In computer vision based fatigue monitoring, there are still some unresolved issues remained, including low re-cognition accuracy in yawn detection based on a single-frame; poor adaptability in blink analysis because of the required threshold, the inability to monitor the transition stages of fatigue in real-time. Attempted to solve these problems, we propose a new classification model in this paper, which is based on two feature time-series for real-time mental fatigue monitoring. First, the mouth opening degree and iris circularity ratio are calculated through facial visual feature extraction. Based on this, we can generate a corresponding time-series called (the proportion of the time during which mouth opening exceeds a given threshold)time series and eye blink time (EBT) time series. Then, using sliding window to partition and annotate the two kinds of time series and build hidden markov model (HMM) for EBT time series. Finally, add a time stamp on HMM to adaptively calculate the initial time point of the next time series, in addition, we can use it to perform the synchronization and fusion of the two time series. Experimental results show that the promoted model can improve yawn detection rate, have good adaptability for blink features of different age groups, and can monitor the transition stage of mental fatigue in real-time.关键词：real time mental fatigue monitoring;iris circularity ratio;mouth opening degree;time series;sliding window;hidden Markov model (HMM)3231|201|0更新时间：2024-05-07
- 摘要：Images are characterized by some statistics of coded vectors in the Bag-of-Features (BOF)model, and then classified by support vector machine (SVM)based on traditional kernel, the existing problems are the loss of discriminant information and choosing of optimal kernel. To solve these problems, we use the multinomial distribution of hard coded vectors or Dirichlet distribution of soft coded vectors as the description of images, and then use maximum likelihood algorithm to estimate the density parameters. Next, the kernel functions between any two images are calculated using a probability product kernel function. Finally, the images are classified by a support vector machine. The experimental results in public image datasets show the proposed algorithm in this paper has achieved better classification performances.关键词：Bag-of-Features model;statistics;probability product kernels;support vector machine;image classification2678|263|0更新时间：2024-05-07
- 摘要：Image classification is an important research field of computer vision, and the key problem of which is to select a type of feature and establish the similarity metrics between images. In view of the mural image characteristics, the contour feature plays an important role in expressing the mural image semantics. Many studies have shown that the contours can be used as an important feature in image recognition and classification. However, previous studies tend to use the chamfer distance between each pair of the most similar contours to compute the similarity between images, or build local descriptors for each contour, clustering into codebook, and describe the image features as histograms, then do the classification using SVM. However, these methods ignore the overall structure between the contours, lack of the overall view of the all contours, while in reality the semantics of an image tend to be more of a holistic semantic. In this paper, we study the similarity metrics between images based on the overall structure of contours, the calculation of contour similarity is subject to the constraint of the space structure relations with other contours, the generated similarity are more able to express the overall similarity between two images. The experimental results show that our method improved accuracy compared to others in mural image classification.关键词：contour features;overall structure;similarity;image classification2508|360|0更新时间：2024-05-07
- 摘要：It is very important to create the background model for image processing.Traditional modeling methods typically use only the gray space,ignoring information from the color space.They do not apply to the complex transportation system.In order to obtain dynamic vehicle information and colored road information for further analysis, we propose a non-parametric road modeling approach based on color spatial clustering in this article. The approach is developed according to the characteristics of pixel distribution in traffic scenes. The modeling process is abstracted as a color spatial clustering process in the time axis. According to the color distortion theory, we define a cluster interval as a cylinder whose axis is the attachment of the clustering center and the origin in color space. Meanwhile, for each pixel the clustering center number is selected adaptively according to the scene’s complexity and change frequency. In this way, the precision improves and the detection is efficient.关键词：intelligent transportation system (ITS);roads modeling;color distortion clustering;RGB color space;color information2784|246|0更新时间：2024-05-07
- 摘要：A spatio-temporal interest point detection approach in the environment of clutter backgrounds with camera shaking/movements is proposed. It incorporates smoothing functions in the spatial and temporal domain. Non-linear anisotropic diffusion is used in spatial direction to efficiently smooth noises and small structures in cluttered backgrounds and to preserve edge information in human action. The uncertainty of the temporal correlation due to camera shaking/movements is modeled by a suppressing function in proportion to the image gradients. This method efficiently solves the problem of cluttered backgrounds and camera shaking/movements in complex scenes.关键词：spatio-temporal interest point detection;nonlinear anisotropic diffusion;clutter background;camera movement2776|225|0更新时间：2024-05-07
- 摘要：Image retrieval based on multiple visual features is one of most important direction in the field of content-based image retrieval. Existing retrieval algorithms are based on multiple features mainly combining features through linear weighting. However, this just achieves combinations in the sense of algebra, and the retrieval result is susceptible to the choice of the weights. To solve this problem, including the extraction of shape and color information in the feature extraction procedure, we propose a method to construct combined shape-color features, called shape-color moment invariants, which are invariant to both shape affine transformation and color diagonal-offset transformations. First, we define the shape core and color core in the two-dimensionol geometric space and three-dimensionol color space respectively. Then multiple integrals to the product of shape core and color core are produced. Finally, we get an invariant feature by normalization. Theoretically, this method can derive an infinite number of invariant features while choosing different shape cores or color cores. The experimental results show that this method is distinctive to existing linear weighting method, and has better performance and efficiency when retrieving near-duplicated images,images with similar property,broadly similar images etc.关键词：shape;color;combined feature;invariants;image retrieval2749|289|0更新时间：2024-05-07
- 摘要：Existing measurements cannot capture global and local features to get the best view of a 3D model. In this paper, we address the problem, and propose a multi-stage method by combining example-learning and edge feature of views. The whole algorithm mainly consists of the following steps. First, Adaboost is applied to select candidate views of the input 3D model by statistical classification and shape similarity. Second, edge information of these views is extracted to define the entropy. It can effectively measure how the candidate views capture local features. Finally, the best viewpoint is selected using a weighted combination of shape similarity and entropy. In our experiments, the algorithm is verified on a 3D model benchmark.关键词：best view;statistical learning;shape feature;edge entropy2440|238|0更新时间：2024-05-07
- 摘要：Due to the intensity inhomogeneity and noise in brain MR images, it is difficult for the traditional models to obtain desirable segmentation results. In this paper, we first propose a local energy function based on the fuzzy C-means model (FCM), which combines segmentation with bias correction. As a result, the proposed model can handle intensity inhomogeneity. Then, the non-local method is used as a regularization term to reduce the impact of noise and to keep the image structure. Finally, the local entropy information is incorporated into the model, which makes it more robust to noise and intensity inhomogeneity. Experiments of the brain magnetic resonance images show that the proposed method can obtain better segmentation results and bias corrected results.关键词：magnetic resonance image;image segmentation;local entropy;non-local spatial information;bias field3067|224|0更新时间：2024-05-07
- 摘要：Accurately and reliably automated segmentation of pulmonary tumors could play an important role in lung cancer diagnosis and radiation oncology. However, it remains a very difficult task in particular for segmenting pulmonary tumors associated with sub-solid nodules that are partially obscured in lung CT images. In this study, we propose and test an improved weighed kernel fuzzy C-means (IWKFCM) method that incorporates vessels structure information and classes’ distribution as weights to segment sub-solid pulmonary nodules. For this purpose, a region of interest (ROI) of a nodule in center CT slice is manually defined. The IWKFCM algorithm is applied to identify and cluster the potential nodule pixels located in this manually-defined center slice and its adjacent (surrounding) slices. The sub-solid nodule is then segmented and defined through 3D connected component labeling and morphological post-processing. This segmentation method is tested using two datasets including 36 nodules selected from a public dataset (LIDC) and 18 nodules depicted on CT images collected from our local hospital. The average overlap ratios between the automated and radiologists’ segmentation of nodules of two datasets are 76.18% and 71.65% respectively. In both datasets, the false-positive ratio (FPR) and false-negative ratio (FNR) are smaller than 17%. Experimental results show that the proposed method enables us to achieve more accurate result in segmenting sub-solid pulmonary nodules than the other previously reported clustering methods. The segmentation results could also provide a consultative reference for more accurately extracting image features and optimal classification of pulmonary nodules in developing computer-aided detection (CAD) schemes.关键词：CAD;improved weighed kernel fuzzy C-means;sub-solid pulmonary nodule;3D segmentation3069|356|0更新时间：2024-05-07
- 摘要：In recent years,the segmentation of coastal hyperspectral remote sensing image which combines ground objects identification with anomaly target detection, is receiving more attention. Moreover, it is a challenge for the massive data and complex construction of hyperspectral image. This paper presents a partial differential equation for active contour models based on the spectral characteristic of coastal terraqueous region segmentation for reference. First, the pixel spectral information of coastal hyperspectral remote sensing image is used as the reference point to construct an energy deviation matrix. Then, using this energy deviation matrix, a new active contour model of terraqueous region segmentation is proposed. It improves the capture and anti-noise capacities of terraqueous marginal region, based on the edge guide function of gradient. The experimental results, compared with that of traditional active contour models, demonstrate that the proposed approach not only yields accurate segmentation but enhances the calculation speed as well.关键词：hyperspectral remote sensing image;coastal zone remote sensing;energy deviation matrix;active contour models;partial differential equation2769|337|0更新时间：2024-05-07
- 摘要：The Bayesian non-local means filter is adapted to the polarimetric synthetic aperture radar (SAR) multiplicative speckle noise, which effectively reduces the speckle noise while keeping the details. Through analyzing the speckle statistics of multi-look SAR images combined with Bayesian non-local means filter, we conclusde that the similarity function in the multi-look SAR image is according to that of the single-look SAR image, which is proved in this paper. Then, a new amended similarity function is proposed. Finally, an adaptive filtering parameter selection method is proposed based on similarity between pels to solve the problem that a fixed parameter could influence the filter performance. The proposed method is validated on real polarimetric SAR images through comparisons with the original algorithm.关键词：polarimetric synthetic aperture radar (SAR);despeckling;non-local means;similarity function;filtering parameter3117|273|0更新时间：2024-05-07
- 摘要：A remote sensing image fusion algorithm is presented based on wavelet transform and sparse representation for the fusion of multi-spectral image and panchromatic image. The algorithm makes full use of the wavelet transform which has the advantage of maintaining spectral information. First, the intensity-hue-saturation (IHS) transform is applied to the multi-spectral image. Then, the obtained corresponding high-and low-frequency coefficients are transformed by the monolayer wavelet transform on the intensity component and the panchromatic image. According to the different characteristics of the low and high frequency coefficients, the low-frequency coefficients cannot be considered to be "sparse". The low-frequency images are obtained their sparse coefficients through sparse representation. The high-frequency coefficients can be considered to be "sparse". A fusion rule, which uses the image information, was taken to compute the high fusion coefficients. Finally, the fused results are obtained through wavelet inverse transform and IHS inverse transform. The experimental results prove that the proposed method improve the spatial resolution and better maintain the spectral characteristics.关键词：image process;image fusion;sparse representation;wavelet transform4137|346|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