Recently, many image encryption algorithms based on chaos have been proposed. Most of the previous algorithms encrypt components R, G, and B of color images independently and neglect the high correlation between them....Recently, many image encryption algorithms based on chaos have been proposed. Most of the previous algorithms encrypt components R, G, and B of color images independently and neglect the high correlation between them. In the paper, a novel color image encryption algorithm is introduced. The 24 bit planes of components R, G, and B of the color plain image are obtained and recombined into 4 compound bit planes, and this can make the three components affect each other. A four-dimensional(4D) memristive hyperchaotic system generates the pseudorandom key streams and its initial values come from the SHA 256 hash value of the color plain image. The compound bit planes and key streams are confused according to the principles of genetic recombination, then confusion and diffusion as a union are applied to the bit planes,and the color cipher image is obtained. Experimental results and security analyses demonstrate that the proposed algorithm is secure and effective so that it may be adopted for secure communication.展开更多
In this paper,a technique based on image pyramid and Bayes rule for reducing noise effects in unsupervised change detection is proposed.By using Gaussian pyramid to process two multitemporal images respectively,two im...In this paper,a technique based on image pyramid and Bayes rule for reducing noise effects in unsupervised change detection is proposed.By using Gaussian pyramid to process two multitemporal images respectively,two image pyramids are constructed.The difference pyramid images are obtained by point-by-point subtraction between the same level images of the two image pyramids.By resizing all difference pyramid images to the size of the original multitemporal image and then making product operator among them,a map being similar to the difference image is obtained.The difference image is generated by point-by-point subtraction between the two multitemporal images directly.At last,the Bayes rule is used to distinguish the changed pixels.Both synthetic and real data sets are used to evaluate the performance of the proposed technique.Experimental results show that the map from the proposed technique is more robust to noise than the difference image.展开更多
A semiautomatic segmentation method based on active contour is proposed for computed tomography (CT) image series. First, to get initial contour, one image slice was segmented exactly by C-V method based on Mumford-Sh...A semiautomatic segmentation method based on active contour is proposed for computed tomography (CT) image series. First, to get initial contour, one image slice was segmented exactly by C-V method based on Mumford-Shah model. Next, the computer will segment the nearby slice automatically using the snake model one by one. During segmenting of image slices, former slice boundary, as next slice initial contour, may cross over next slice real boundary and never return to right position. To avoid contour skipping over, the distance variance between two slices is evaluated by an threshold, which decides whether to initiate again. Moreover, a new improved marching cubes (MC) algorithm based on 2D images series segmentation boundary is given for 3D image reconstruction. Compared with the standard method, the proposed algorithm reduces detecting time and needs less storing memory. The effectiveness and capabilities of the algorithm were illustrated by experimental results.展开更多
At present, many chaos-based image encryption algorithms have proved to be unsafe, few encryption schemes permute the plain images as three-dimensional（3D） bit matrices, and thus bits cannot move to any position, th...At present, many chaos-based image encryption algorithms have proved to be unsafe, few encryption schemes permute the plain images as three-dimensional（3D） bit matrices, and thus bits cannot move to any position, the movement range of bits are limited, and based on them, in this paper we present a novel image encryption algorithm based on 3D Brownian motion and chaotic systems. The architecture of confusion and diffusion is adopted. Firstly, the plain image is converted into a 3D bit matrix and split into sub blocks. Secondly, block confusion based on 3D Brownian motion（BCB3DBM）is proposed to permute the position of the bits within the sub blocks, and the direction of particle movement is generated by logistic-tent system（LTS）. Furthermore, block confusion based on position sequence group（BCBPSG） is introduced, a four-order memristive chaotic system is utilized to give random chaotic sequences, and the chaotic sequences are sorted and a position sequence group is chosen based on the plain image, then the sub blocks are confused. The proposed confusion strategy can change the positions of the bits and modify their weights, and effectively improve the statistical performance of the algorithm. Finally, a pixel level confusion is employed to enhance the encryption effect. The initial values and parameters of chaotic systems are produced by the SHA 256 hash function of the plain image. Simulation results and security analyses illustrate that our algorithm has excellent encryption performance in terms of security and speed.展开更多
This paper addresses the application of hand gesture recognition in monocular image sequences using Active Appearance Model （AAM）, For this work, the proposed algorithm is composed of constricting AAMs and fitting t...This paper addresses the application of hand gesture recognition in monocular image sequences using Active Appearance Model （AAM）, For this work, the proposed algorithm is composed of constricting AAMs and fitting the models to the interest region. In training stage, according to the manual labeled feature points, the relative AAM is constructed and the corresponding average feature is obtained. In recognition stage, the interesting hand gesture region is firstly segmented by skin and movement cues. Secondly, the models are fitted to the image that includes the hand gesture, and the relative features are extracted. Thirdly, the classification is done by comparing the extracted features and average features. 30 different gestures of Chinese sign language are applied for testing the effectiveness of the method. The Experimental results are given indicating good performance of the algorithm.展开更多
The trained Gaussian mixture model is used to make skincolour segmentation for the input image sequences. The hand gesture region is extracted, and the relative normalization images are obtained by interpolation opera...The trained Gaussian mixture model is used to make skincolour segmentation for the input image sequences. The hand gesture region is extracted, and the relative normalization images are obtained by interpolation operation. To solve the proem of hand gesture recognition, Fuzzy-Rough based nearest neighbour（RNN） algorithm is applied for classification. For avoiding the costly compute, an improved nearest neighbour classification algorithm based on fuzzy-rough set theory （FRNNC） is proposed. The algorithm employs the represented cluster points instead of the whole training samples, and takes the hand gesture data＇s fuzziness and the roughness into account, so the campute spending is decreased and the recognition rate is increased. The 30 gestures in Chinese sign language alphabet are used for approving the effectiveness of the proposed algorithm. The recognition rate is 94.96%, which is better than that of KNN （K nearest neighbor）and Fuzzy- KNN （Fuzzy K nearest neighbor）.展开更多
A new trend in the development of medical image processing systems is to enhance the shar-ing of medical resources and the collaborative processing of medical specialists. This paper presents an architecture of medica...A new trend in the development of medical image processing systems is to enhance the shar-ing of medical resources and the collaborative processing of medical specialists. This paper presents an architecture of medical image dynamic collaborative processing on the distributed environment by combining the JAVA, CORBA (Common Object Request and Broker Architecture) and the MAS (Multi-Agents System) collaborative mechanism. The architecture allows medical specialists or applications to share records and cornmunicate with each other on the web by overcoming the shortcut of traditional approach using Common Gateway Interface (CGI) and client/server architecture, and can support the remote heterogeneous systems collaboration. The new approach im-proves the collaborative processing of medical data and applications and is able to enhance the in-teroperation among heterogeneous system. Research on the system will help the collaboration and cooperation among medical application systems distributed on the web, thus supply high quality medical service such as diagnosis and therapy to practicing specialists regardless of their actual geo-graphic location.展开更多
Multi-modal histological image registration tasks pose significant challenges due to tissue staining operations causing partial loss and folding of tissue.Convolutional neural network(CNN)and generative adversarial ne...Multi-modal histological image registration tasks pose significant challenges due to tissue staining operations causing partial loss and folding of tissue.Convolutional neural network(CNN)and generative adversarial network(GAN)are pivotal inmedical image registration.However,existing methods often struggle with severe interference and deformation,as seen in histological images of conditions like Cushing’s disease.We argue that the failure of current approaches lies in underutilizing the feature extraction capability of the discriminator inGAN.In this study,we propose a novel multi-modal registration approach GAN-DIRNet based on GAN for deformable histological image registration.To begin with,the discriminators of two GANs are embedded as a new dual parallel feature extraction module into the unsupervised registration networks,characterized by implicitly extracting feature descriptors of specific modalities.Additionally,modal feature description layers and registration layers collaborate in unsupervised optimization,facilitating faster convergence and more precise results.Lastly,experiments and evaluations were conducted on the registration of the Mixed National Institute of Standards and Technology database(MNIST),eight publicly available datasets of histological sections and the Clustering-Registration-Classification-Segmentation(CRCS)dataset on the Cushing’s disease.Experimental results demonstrate that our proposed GAN-DIRNet method surpasses existing approaches like DIRNet in terms of both registration accuracy and time efficiency,while also exhibiting robustness across different image types.展开更多
Diagnostic ultrasound is a useful and noninvasive method in clinical medicine. Although due to its qualitative, sub- jective and experience-based nature, ultrasound image interpretation can be influenced by image cond...Diagnostic ultrasound is a useful and noninvasive method in clinical medicine. Although due to its qualitative, sub- jective and experience-based nature, ultrasound image interpretation can be influenced by image conditions such as scanning frequency and machine settings. In this paper, a novel method is proposed to extract the liver features using the joint features of fractal dimension and the entropies of texture edge co-occurrence matrix based on ultrasound images, which is not sensitive to changes in emission frequency and gain. Then, Fisher linear classifier and support vector machine are employed to test a group of 99 in-vivo liver fibrosis images from 18 patients, as well as other 273 liver images from 18 normal human volunteers.展开更多
Face recognition has attracted great interest due to its importance in many real-world applications. In this paper,we present a novel low-rank sparse representation-based classification(LRSRC) method for robust face r...Face recognition has attracted great interest due to its importance in many real-world applications. In this paper,we present a novel low-rank sparse representation-based classification(LRSRC) method for robust face recognition. Given a set of test samples, LRSRC seeks the lowest-rank and sparsest representation matrix over all training samples. Since low-rank model can reveal the subspace structures of data while sparsity helps to recognize the data class, the obtained test sample representations are both representative and discriminative. Using the representation vector of a test sample, LRSRC classifies the test sample into the class which generates minimal reconstruction error. Experimental results on several face image databases show the effectiveness and robustness of LRSRC in face image recognition.展开更多
Segmentation of the bladder in computerized tomography(CT) images is an important step in radiation therapy planning of prostate cancer. We present a new segmentation scheme to automatically delineate the bladder cont...Segmentation of the bladder in computerized tomography(CT) images is an important step in radiation therapy planning of prostate cancer. We present a new segmentation scheme to automatically delineate the bladder contour in CT images with three major steps. First,we use the mean shift algorithm to obtain a clustered image containing the rough contour of the bladder,which is then extracted in the second step by applying a region-growing algorithm with the initial seed point selected from a line-by-line scanning process. The third step is to refine the bladder contour more accurately using the rolling-ball algorithm. These steps are then extended to segment the bladder volume in a slice-by-slice manner. The obtained results were compared to manual segmentation by radiation oncologists. The average values of sensitivity,specificity,positive predictive value,negative predictive value,and Hausdorff distance are 86.5%,96.3%,90.5%,96.5%,and 2.8 pixels,respectively. The results show that the bladder can be accurately segmented.展开更多
This paper attempts to introduce an improved difference model that modifies a car-following model, which takes the next-nearest-neighbor interaction into account. The hnprovement of this model over the previous one li...This paper attempts to introduce an improved difference model that modifies a car-following model, which takes the next-nearest-neighbor interaction into account. The hnprovement of this model over the previous one lies in that it performs more realistically in the dynamical motion for small delay time. The traffic behavior of the improved model is investigated with analytic and numerical methods with the finding that the new consideration could further stabilize traffic flow. And some simulation tests verify that the proposed model can demonstrate some complex physical features observed recently in real traffic such as the existence of three phases： free flow, coexisting flow, and jam flow; spontaneous formation of density waves; sudden flow drop in flow-density plane; traffic hysteresis in transition between the free and the coexisting flow. Furthermore, th.e improved model also predicts that the stable state to relative density in the coexisting flow is insusceptible to noise.展开更多
A support vector regression(SVR) based color image restoration algorithm is proposed.The test color images are firstly mapped into the YUV color space,and then SVR is applied to build up a theoretical model between th...A support vector regression(SVR) based color image restoration algorithm is proposed.The test color images are firstly mapped into the YUV color space,and then SVR is applied to build up a theoretical model between the degraded images and the original one.Performance comparisons of the proposed algorithm versus traditional filtering algorithms are given.Experimental results show that the proposed algorithm has better performance than traditional filtering algorithms and has less computation time than iterative blind deconvolution algorithm.展开更多
Effective and robust recognition and tracking of objects are the key problems in visual surveillance systems. Most existing object recognition methods were designed with particular objects in mind. This study presents...Effective and robust recognition and tracking of objects are the key problems in visual surveillance systems. Most existing object recognition methods were designed with particular objects in mind. This study presents a general moving objects recognition method using global features of targets. Targets are extracted with an adaptive Gaussian mixture model and their silhouette images are captured and unified. A new objects silhouette database is built to provide abundant samples to train the subspace feature. This database is more convincing than the previous ones. A more effective dimension reduction method based on graph embedding is used to obtain the projection eigenvector. In our experiments, we show the effective performance of our method in addressing the moving objects recognition problem and its superiority compared with the previous methods.展开更多
In this paper, we introduce a novel class of coplanar conics, the pencil of which can doubly contact to calibrate camera and estimate pose. We first analyze the properties of con-axes and con-eccentricity ellipses, wh...In this paper, we introduce a novel class of coplanar conics, the pencil of which can doubly contact to calibrate camera and estimate pose. We first analyze the properties of con-axes and con-eccentricity ellipses, which consist of a naturM extending pattern of concentric circles. Then the general case that two ellipses have two repeated complex intersection points is presented. This degenerate configuration results in a one-parameter family of homographies which map the planar pattern to its image. Although it is unable to compute the complete homography, an indirect 3-degree polynomial or 5-degree polynomial constraint on intrinsic parameters from one image can also be used for camera calibration and pose estimation under the minimal conditions. Furthermore, this nonlinear problem can be treated as a polynomial optimization problem （POP） and the global optimization solution can be also obtained by using SparsePOP （a sparse semidefinite programming relaxation of POPs）, Finally, the experiments with simulated data and real images are shown to verify the correctness and robustness of the proposed technique.展开更多
Taking into account the demands of hyperspectral remote sensing(RS) image retrieval and processing, some encoding methods of spectral vector including direct encoding, feature-based encoding and tree-based encoding me...Taking into account the demands of hyperspectral remote sensing(RS) image retrieval and processing, some encoding methods of spectral vector including direct encoding, feature-based encoding and tree-based encoding methods are proposed and compared. In direct encoding, based on the analysis of binary encoding and quad-value encoding, decimal encoding is proposed. It is proved that quad-value encoding and decimal encoding are suitable to fast processing and retrieval. In absorption feature-based encoding method, five common metrics are compared. Because locations of reflection/absorption features are sensitive to noise, this method is not very effective in retrieval. In tree-based encoding methods, bitree, quadtree, octree and hextree are proposed and discussed. It is proved that 2-level octree and 2-level hextree are more effective than bitree and quadtree. Finally, quad-value encoding, decimal encoding, 2-level octree and 2-level hextree are proposed in spectral vectors encoding, similarity measure and hyperspectral RS image retrieval.展开更多
The generative adversarial network(GAN)is first proposed in 2014,and this kind of network model is machine learning systems that can learn to measure a given distribution of data,one of the most important applications...The generative adversarial network(GAN)is first proposed in 2014,and this kind of network model is machine learning systems that can learn to measure a given distribution of data,one of the most important applications is style transfer.Style transfer is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image.CYCLE-GAN is a classic GAN model,which has a wide range of scenarios in style transfer.Considering its unsupervised learning characteristics,the mapping is easy to be learned between an input image and an output image.However,it is difficult for CYCLE-GAN to converge and generate high-quality images.In order to solve this problem,spectral normalization is introduced into each convolutional kernel of the discriminator.Every convolutional kernel reaches Lipschitz stability constraint with adding spectral normalization and the value of the convolutional kernel is limited to[0,1],which promotes the training process of the proposed model.Besides,we use pretrained model(VGG16)to control the loss of image content in the position of l1 regularization.To avoid overfitting,l1 regularization term and l2 regularization term are both used in the object loss function.In terms of Frechet Inception Distance(FID)score evaluation,our proposed model achieves outstanding performance and preserves more discriminative features.Experimental results show that the proposed model converges faster and achieves better FID scores than the state of the art.展开更多
Automatic recognition of artists is very important in acoustic music indexing, browsing, and contentbased acoustic music retrieving, but synchronously it is still a challenging errand to extract the most representativ...Automatic recognition of artists is very important in acoustic music indexing, browsing, and contentbased acoustic music retrieving, but synchronously it is still a challenging errand to extract the most representative and salient attributes to depict diversiform artists. In this paper, we developed a novel system to complete the reorganization of artist automatically. The proposed system can efficiently identify the artist's voice of a raw song by analyzing substantive features extracted from both pure music and singing song mixed with accompanying music. The experiments on different genres of songs illustrate that the proposed system is possible.展开更多
Linear discrimiant analysis (LDA) has been used in face recognition. But it is difficult to handle the high nonlinear problems, such as changes of large viewpoint and illumination. In order to overcome these problems,...Linear discrimiant analysis (LDA) has been used in face recognition. But it is difficult to handle the high nonlinear problems, such as changes of large viewpoint and illumination. In order to overcome these problems, kernel discriminant analysis for face recognition is presented. This approach adopts the kernel functions to replace the dot products of nonlinear mapping in the high dimensional feature space, and then the nonlinear problem can be solved in the input space conveniently without explicit mapping. Two face databases are given.展开更多
A novel approach based on independent component analysis (ICA) for speckle filtering and target extraction of synthetic aperture radar (SAR) images is proposed using adaptive space separation with weighted information...A novel approach based on independent component analysis (ICA) for speckle filtering and target extraction of synthetic aperture radar (SAR) images is proposed using adaptive space separation with weighted information entropy (WIE) incorporated. First the basis and the independent components are respectively obtained by ICA technique, and WIE of the image is computed; then based on the threshold computed from function T-WIE (threshold versus weighted-information-entropy), independent components are adaptively separated and the bases are classified accordingly. Thus, the image space is separated into two subspaces: "clean" and "noise". Then, a proposed nonlinear operator ABO is applied on each component of the 'clean' subspace for further optimization. Finally, recovery image is obtained reconstructing this subspace and target is easily extracted with binarisation. Note that here T-WIE is an interpolated function based on several representative target SAR images using proposed space separation algorithm.展开更多
基金supported by the National Natural Science Foundation of China(Grant Nos.61203094 and 61305042)the Natural Science Foundation of the United States(Grant Nos.CNS-1253424 and ECCS-1202225)+3 种基金the Science and Technology Foundation of Henan Province,China(Grant No.152102210048)the Foundation and Frontier Project of Henan Province,China(Grant No.162300410196)the Natural Science Foundation of Educational Committee of Henan Province,China(Grant No.14A413015)the Research Foundation of Henan University,China(Grant No.xxjc20140006)
文摘Recently, many image encryption algorithms based on chaos have been proposed. Most of the previous algorithms encrypt components R, G, and B of color images independently and neglect the high correlation between them. In the paper, a novel color image encryption algorithm is introduced. The 24 bit planes of components R, G, and B of the color plain image are obtained and recombined into 4 compound bit planes, and this can make the three components affect each other. A four-dimensional(4D) memristive hyperchaotic system generates the pseudorandom key streams and its initial values come from the SHA 256 hash value of the color plain image. The compound bit planes and key streams are confused according to the principles of genetic recombination, then confusion and diffusion as a union are applied to the bit planes,and the color cipher image is obtained. Experimental results and security analyses demonstrate that the proposed algorithm is secure and effective so that it may be adopted for secure communication.
基金the National Basic Research Program(973) of China (No. 2006CB701303)the National High Technology Research and Development Program(863) of China (No. 2006AA12Z105)
文摘In this paper,a technique based on image pyramid and Bayes rule for reducing noise effects in unsupervised change detection is proposed.By using Gaussian pyramid to process two multitemporal images respectively,two image pyramids are constructed.The difference pyramid images are obtained by point-by-point subtraction between the same level images of the two image pyramids.By resizing all difference pyramid images to the size of the original multitemporal image and then making product operator among them,a map being similar to the difference image is obtained.The difference image is generated by point-by-point subtraction between the two multitemporal images directly.At last,the Bayes rule is used to distinguish the changed pixels.Both synthetic and real data sets are used to evaluate the performance of the proposed technique.Experimental results show that the map from the proposed technique is more robust to noise than the difference image.
文摘A semiautomatic segmentation method based on active contour is proposed for computed tomography (CT) image series. First, to get initial contour, one image slice was segmented exactly by C-V method based on Mumford-Shah model. Next, the computer will segment the nearby slice automatically using the snake model one by one. During segmenting of image slices, former slice boundary, as next slice initial contour, may cross over next slice real boundary and never return to right position. To avoid contour skipping over, the distance variance between two slices is evaluated by an threshold, which decides whether to initiate again. Moreover, a new improved marching cubes (MC) algorithm based on 2D images series segmentation boundary is given for 3D image reconstruction. Compared with the standard method, the proposed algorithm reduces detecting time and needs less storing memory. The effectiveness and capabilities of the algorithm were illustrated by experimental results.
基金Project supported by the National Natural Science Foundation of China(Grant Nos.41571417 and 61305042)the National Science Foundation of the United States(Grant Nos.CNS-1253424 and ECCS-1202225)+4 种基金the Science and Technology Foundation of Henan Province,China(Grant No.152102210048)the Foundation and Frontier Project of Henan Province,China(Grant No.162300410196)China Postdoctoral Science Foundation(Grant No.2016M602235)the Natural Science Foundation of Educational Committee of Henan Province,China(Grant No.14A413015)the Research Foundation of Henan University,China(Grant No.xxjc20140006)
文摘At present, many chaos-based image encryption algorithms have proved to be unsafe, few encryption schemes permute the plain images as three-dimensional（3D） bit matrices, and thus bits cannot move to any position, the movement range of bits are limited, and based on them, in this paper we present a novel image encryption algorithm based on 3D Brownian motion and chaotic systems. The architecture of confusion and diffusion is adopted. Firstly, the plain image is converted into a 3D bit matrix and split into sub blocks. Secondly, block confusion based on 3D Brownian motion（BCB3DBM）is proposed to permute the position of the bits within the sub blocks, and the direction of particle movement is generated by logistic-tent system（LTS）. Furthermore, block confusion based on position sequence group（BCBPSG） is introduced, a four-order memristive chaotic system is utilized to give random chaotic sequences, and the chaotic sequences are sorted and a position sequence group is chosen based on the plain image, then the sub blocks are confused. The proposed confusion strategy can change the positions of the bits and modify their weights, and effectively improve the statistical performance of the algorithm. Finally, a pixel level confusion is employed to enhance the encryption effect. The initial values and parameters of chaotic systems are produced by the SHA 256 hash function of the plain image. Simulation results and security analyses illustrate that our algorithm has excellent encryption performance in terms of security and speed.
文摘This paper addresses the application of hand gesture recognition in monocular image sequences using Active Appearance Model （AAM）, For this work, the proposed algorithm is composed of constricting AAMs and fitting the models to the interest region. In training stage, according to the manual labeled feature points, the relative AAM is constructed and the corresponding average feature is obtained. In recognition stage, the interesting hand gesture region is firstly segmented by skin and movement cues. Secondly, the models are fitted to the image that includes the hand gesture, and the relative features are extracted. Thirdly, the classification is done by comparing the extracted features and average features. 30 different gestures of Chinese sign language are applied for testing the effectiveness of the method. The Experimental results are given indicating good performance of the algorithm.
文摘The trained Gaussian mixture model is used to make skincolour segmentation for the input image sequences. The hand gesture region is extracted, and the relative normalization images are obtained by interpolation operation. To solve the proem of hand gesture recognition, Fuzzy-Rough based nearest neighbour（RNN） algorithm is applied for classification. For avoiding the costly compute, an improved nearest neighbour classification algorithm based on fuzzy-rough set theory （FRNNC） is proposed. The algorithm employs the represented cluster points instead of the whole training samples, and takes the hand gesture data＇s fuzziness and the roughness into account, so the campute spending is decreased and the recognition rate is increased. The 30 gestures in Chinese sign language alphabet are used for approving the effectiveness of the proposed algorithm. The recognition rate is 94.96%, which is better than that of KNN （K nearest neighbor）and Fuzzy- KNN （Fuzzy K nearest neighbor）.
基金the National Nature Science Foundation of China.
文摘A new trend in the development of medical image processing systems is to enhance the shar-ing of medical resources and the collaborative processing of medical specialists. This paper presents an architecture of medical image dynamic collaborative processing on the distributed environment by combining the JAVA, CORBA (Common Object Request and Broker Architecture) and the MAS (Multi-Agents System) collaborative mechanism. The architecture allows medical specialists or applications to share records and cornmunicate with each other on the web by overcoming the shortcut of traditional approach using Common Gateway Interface (CGI) and client/server architecture, and can support the remote heterogeneous systems collaboration. The new approach im-proves the collaborative processing of medical data and applications and is able to enhance the in-teroperation among heterogeneous system. Research on the system will help the collaboration and cooperation among medical application systems distributed on the web, thus supply high quality medical service such as diagnosis and therapy to practicing specialists regardless of their actual geo-graphic location.
文摘Multi-modal histological image registration tasks pose significant challenges due to tissue staining operations causing partial loss and folding of tissue.Convolutional neural network(CNN)and generative adversarial network(GAN)are pivotal inmedical image registration.However,existing methods often struggle with severe interference and deformation,as seen in histological images of conditions like Cushing’s disease.We argue that the failure of current approaches lies in underutilizing the feature extraction capability of the discriminator inGAN.In this study,we propose a novel multi-modal registration approach GAN-DIRNet based on GAN for deformable histological image registration.To begin with,the discriminators of two GANs are embedded as a new dual parallel feature extraction module into the unsupervised registration networks,characterized by implicitly extracting feature descriptors of specific modalities.Additionally,modal feature description layers and registration layers collaborate in unsupervised optimization,facilitating faster convergence and more precise results.Lastly,experiments and evaluations were conducted on the registration of the Mixed National Institute of Standards and Technology database(MNIST),eight publicly available datasets of histological sections and the Clustering-Registration-Classification-Segmentation(CRCS)dataset on the Cushing’s disease.Experimental results demonstrate that our proposed GAN-DIRNet method surpasses existing approaches like DIRNet in terms of both registration accuracy and time efficiency,while also exhibiting robustness across different image types.
文摘Diagnostic ultrasound is a useful and noninvasive method in clinical medicine. Although due to its qualitative, sub- jective and experience-based nature, ultrasound image interpretation can be influenced by image conditions such as scanning frequency and machine settings. In this paper, a novel method is proposed to extract the liver features using the joint features of fractal dimension and the entropies of texture edge co-occurrence matrix based on ultrasound images, which is not sensitive to changes in emission frequency and gain. Then, Fisher linear classifier and support vector machine are employed to test a group of 99 in-vivo liver fibrosis images from 18 patients, as well as other 273 liver images from 18 normal human volunteers.
基金supported by National Natural Science Foundation of China(No.61374134)the key Scientific Research Project of Universities in Henan Province,China(No.15A413009)
文摘Face recognition has attracted great interest due to its importance in many real-world applications. In this paper,we present a novel low-rank sparse representation-based classification(LRSRC) method for robust face recognition. Given a set of test samples, LRSRC seeks the lowest-rank and sparsest representation matrix over all training samples. Since low-rank model can reveal the subspace structures of data while sparsity helps to recognize the data class, the obtained test sample representations are both representative and discriminative. Using the representation vector of a test sample, LRSRC classifies the test sample into the class which generates minimal reconstruction error. Experimental results on several face image databases show the effectiveness and robustness of LRSRC in face image recognition.
基金Project (No. 60675023) supported by the National Natural Science Foundation of China
文摘Segmentation of the bladder in computerized tomography(CT) images is an important step in radiation therapy planning of prostate cancer. We present a new segmentation scheme to automatically delineate the bladder contour in CT images with three major steps. First,we use the mean shift algorithm to obtain a clustered image containing the rough contour of the bladder,which is then extracted in the second step by applying a region-growing algorithm with the initial seed point selected from a line-by-line scanning process. The third step is to refine the bladder contour more accurately using the rolling-ball algorithm. These steps are then extended to segment the bladder volume in a slice-by-slice manner. The obtained results were compared to manual segmentation by radiation oncologists. The average values of sensitivity,specificity,positive predictive value,negative predictive value,and Hausdorff distance are 86.5%,96.3%,90.5%,96.5%,and 2.8 pixels,respectively. The results show that the bladder can be accurately segmented.
基金The project supported by the Key Foundation Project of Shanghai under Grant No. 032912066
文摘This paper attempts to introduce an improved difference model that modifies a car-following model, which takes the next-nearest-neighbor interaction into account. The hnprovement of this model over the previous one lies in that it performs more realistically in the dynamical motion for small delay time. The traffic behavior of the improved model is investigated with analytic and numerical methods with the finding that the new consideration could further stabilize traffic flow. And some simulation tests verify that the proposed model can demonstrate some complex physical features observed recently in real traffic such as the existence of three phases： free flow, coexisting flow, and jam flow; spontaneous formation of density waves; sudden flow drop in flow-density plane; traffic hysteresis in transition between the free and the coexisting flow. Furthermore, th.e improved model also predicts that the stable state to relative density in the coexisting flow is insusceptible to noise.
基金the National Natural Science Foundation of China (No. 60675023)
文摘A support vector regression(SVR) based color image restoration algorithm is proposed.The test color images are firstly mapped into the YUV color space,and then SVR is applied to build up a theoretical model between the degraded images and the original one.Performance comparisons of the proposed algorithm versus traditional filtering algorithms are given.Experimental results show that the proposed algorithm has better performance than traditional filtering algorithms and has less computation time than iterative blind deconvolution algorithm.
基金Project (No. 60805001) partially supported by the National NaturalScience Foundation of China
文摘Effective and robust recognition and tracking of objects are the key problems in visual surveillance systems. Most existing object recognition methods were designed with particular objects in mind. This study presents a general moving objects recognition method using global features of targets. Targets are extracted with an adaptive Gaussian mixture model and their silhouette images are captured and unified. A new objects silhouette database is built to provide abundant samples to train the subspace feature. This database is more convincing than the previous ones. A more effective dimension reduction method based on graph embedding is used to obtain the projection eigenvector. In our experiments, we show the effective performance of our method in addressing the moving objects recognition problem and its superiority compared with the previous methods.
基金the National Basic Research Program (973) of China(No.2011CB302203)the National Natural Science Foundation of China(No.60833009)
文摘In this paper, we introduce a novel class of coplanar conics, the pencil of which can doubly contact to calibrate camera and estimate pose. We first analyze the properties of con-axes and con-eccentricity ellipses, which consist of a naturM extending pattern of concentric circles. Then the general case that two ellipses have two repeated complex intersection points is presented. This degenerate configuration results in a one-parameter family of homographies which map the planar pattern to its image. Although it is unable to compute the complete homography, an indirect 3-degree polynomial or 5-degree polynomial constraint on intrinsic parameters from one image can also be used for camera calibration and pose estimation under the minimal conditions. Furthermore, this nonlinear problem can be treated as a polynomial optimization problem （POP） and the global optimization solution can be also obtained by using SparsePOP （a sparse semidefinite programming relaxation of POPs）, Finally, the experiments with simulated data and real images are shown to verify the correctness and robustness of the proposed technique.
文摘Taking into account the demands of hyperspectral remote sensing(RS) image retrieval and processing, some encoding methods of spectral vector including direct encoding, feature-based encoding and tree-based encoding methods are proposed and compared. In direct encoding, based on the analysis of binary encoding and quad-value encoding, decimal encoding is proposed. It is proved that quad-value encoding and decimal encoding are suitable to fast processing and retrieval. In absorption feature-based encoding method, five common metrics are compared. Because locations of reflection/absorption features are sensitive to noise, this method is not very effective in retrieval. In tree-based encoding methods, bitree, quadtree, octree and hextree are proposed and discussed. It is proved that 2-level octree and 2-level hextree are more effective than bitree and quadtree. Finally, quad-value encoding, decimal encoding, 2-level octree and 2-level hextree are proposed in spectral vectors encoding, similarity measure and hyperspectral RS image retrieval.
基金This work is supported by the National Natural Science Foundation of China(No.61702226)the 111 Project(B12018)+1 种基金the Natural Science Foundation of Jiangsu Province(No.BK20170200)the Fundamental Research Funds for the Central Universities(No.JUSRP11854).
文摘The generative adversarial network(GAN)is first proposed in 2014,and this kind of network model is machine learning systems that can learn to measure a given distribution of data,one of the most important applications is style transfer.Style transfer is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image.CYCLE-GAN is a classic GAN model,which has a wide range of scenarios in style transfer.Considering its unsupervised learning characteristics,the mapping is easy to be learned between an input image and an output image.However,it is difficult for CYCLE-GAN to converge and generate high-quality images.In order to solve this problem,spectral normalization is introduced into each convolutional kernel of the discriminator.Every convolutional kernel reaches Lipschitz stability constraint with adding spectral normalization and the value of the convolutional kernel is limited to[0,1],which promotes the training process of the proposed model.Besides,we use pretrained model(VGG16)to control the loss of image content in the position of l1 regularization.To avoid overfitting,l1 regularization term and l2 regularization term are both used in the object loss function.In terms of Frechet Inception Distance(FID)score evaluation,our proposed model achieves outstanding performance and preserves more discriminative features.Experimental results show that the proposed model converges faster and achieves better FID scores than the state of the art.
基金the National Natural Science Foundation of China (No. 60675017)the National Basic Research Program (973) of China (No. 2006CB303103)
文摘Automatic recognition of artists is very important in acoustic music indexing, browsing, and contentbased acoustic music retrieving, but synchronously it is still a challenging errand to extract the most representative and salient attributes to depict diversiform artists. In this paper, we developed a novel system to complete the reorganization of artist automatically. The proposed system can efficiently identify the artist's voice of a raw song by analyzing substantive features extracted from both pure music and singing song mixed with accompanying music. The experiments on different genres of songs illustrate that the proposed system is possible.
文摘Linear discrimiant analysis (LDA) has been used in face recognition. But it is difficult to handle the high nonlinear problems, such as changes of large viewpoint and illumination. In order to overcome these problems, kernel discriminant analysis for face recognition is presented. This approach adopts the kernel functions to replace the dot products of nonlinear mapping in the high dimensional feature space, and then the nonlinear problem can be solved in the input space conveniently without explicit mapping. Two face databases are given.
基金the National Natural Science Foundation of China (No. 60675023)the Aviation Science Foundation of China (No. 04F57004)
文摘A novel approach based on independent component analysis (ICA) for speckle filtering and target extraction of synthetic aperture radar (SAR) images is proposed using adaptive space separation with weighted information entropy (WIE) incorporated. First the basis and the independent components are respectively obtained by ICA technique, and WIE of the image is computed; then based on the threshold computed from function T-WIE (threshold versus weighted-information-entropy), independent components are adaptively separated and the bases are classified accordingly. Thus, the image space is separated into two subspaces: "clean" and "noise". Then, a proposed nonlinear operator ABO is applied on each component of the 'clean' subspace for further optimization. Finally, recovery image is obtained reconstructing this subspace and target is easily extracted with binarisation. Note that here T-WIE is an interpolated function based on several representative target SAR images using proposed space separation algorithm.
