Although scene classification has been studied for decades, indoor scene recognition remains challenging due to its large view point variance and massive irregular artefacts. In fact, most existing methods for outdoor...
详细信息
ISBN:
(纸本)9781424496297
Although scene classification has been studied for decades, indoor scene recognition remains challenging due to its large view point variance and massive irregular artefacts. In fact, most existing methods for outdoor scene classification perform poorly in the indoor situation. To address the problem, we propose a hybrid image representation by combining the global information with the local structure of the scene. First, the global discriminative information is captured by pyramid GIST feature. Second, the local structure is encoded by the bag of features method with Histogram Intersection Kernel (HIK). Finally, HIK based SVM is employed for learning and classification. Experiments on the MIT indoor scene database show that our approach could significantly improve the recognition accuracy of the state-of-art methods by about 14%.
This paper addresses a strategy for 3D human motion recovery from monocular image. We advocate the use of Gaussian Process Dynamical Model (GPDM) for learning human pose and motion priors for 3D people tracking. With ...
详细信息
This paper addresses a strategy for 3D human motion recovery from monocular image. We advocate the use of Gaussian Process Dynamical Model (GPDM) for learning human pose and motion priors for 3D people tracking. With the prior learned from GPDM, we integrate our approach into a Bayesian tracking framework of condensation. During the off-line training step, a GPDM provides the reversible mappings between low-dimensional latent space and high-dimensional pose space, and then in the online tracking process, the latent variables are estimated via the particle filtering, and the observation is designed as a energy function based on a Markov Random Field (MRF) theory. The proposed approach is demonstrated on our database, and the experimental results show that our method performs promisingly.
Conotoxins show prospects for being potent pharmaceuticals in the treatment of some serious disease. Accurate prediction of conotoxin superfamily would have many important applications in biological research and clini...
详细信息
Conotoxins show prospects for being potent pharmaceuticals in the treatment of some serious disease. Accurate prediction of conotoxin superfamily would have many important applications in biological research and clinical medicine. In this study, we propose a novel dHKNN method to predict conotoxin superfamily. Firstly, we extract the protein's sequential features composed of physicochemical properties, evolutionary information, predicted secondary structures and amino acid composition. Then we use the diffusion maps for dimensionality *** last, with considering the local density information in the diffusion space, the dHKNN is proposed based on the K-local hyperplane distance nearest neighbor subspace classifier method for predicting conotoxin superfamilies. An overall accuracy of 91.90% is obtained through the jackknife cross-validation test which is higher than present methods.
De-noising algorithms for volume CT are proposed based on the correlation of the cone beam (CB) CT projections. The main idea is: when the correlated pixels among the neighboring views are added together according to ...
详细信息
De-noising algorithms for volume CT are proposed based on the correlation of the cone beam (CB) CT projections. The main idea is: when the correlated pixels among the neighboring views are added together according to this projection correlation (PC), the signal will be retained, whereas the noise variance upon them will be reduced by this averaging process, because the signal obeys this correlation but noise not. PC based filtering (PCF) method is developed. PCF is further integrated as PC based penalty weighted least square (PC-PWLS) method. Validations are performed for two volume scans: one is ideal low-dose case, and the other considers the high-frequency scatter residues after scatter correction. By using the proposed methods, significant noise suppressing is illustrated; meanwhile, the spatial resolution can be well preserved. For the low dose scan, SNR (Signal to Noise Ratio) increases are around 14 dB. For the cone beam scan with high frequency scatter residues, an effect approaching to the fan-beam scan is achieved.
Saliency mechanism has been considered crucial in the human visual system and helpful to object detection and recognition. This paper addresses a novel feature-based model for visual saliency detection. It consists of...
详细信息
Saliency mechanism has been considered crucial in the human visual system and helpful to object detection and recognition. This paper addresses a novel feature-based model for visual saliency detection. It consists of two steps: first, using the learned overcomplete sparse bases to represent image patches; and then, estimating saliency information via direct low-rank and sparsity matrix decomposition. We compare our model with the previous methods on natural images. Experimental results show that our model performs competitively for visual saliency detection task, and suggest the potential application of matrix decomposition and convex optimization for image analysis.
By introducing a novel membership constraint function, a new algorithm called fuzzy c-means switching regression model with generalized improved fuzzy partitions (GIFP-FCRM) is proposed. This algorithm seems less sens...
详细信息
By introducing a novel membership constraint function, a new algorithm called fuzzy c-means switching regression model with generalized improved fuzzy partitions (GIFP-FCRM) is proposed. This algorithm seems less sensitive to noise and outliers than the classical fuzzy C switching regression model (FCRM), and provides a generalized model with the fuzziness index m for the fuzzy C switching regression model with improved fuzzy partitions (IFP-FCRM). Furthermore, with fuzzy parameter α, the classical FCRM and IFP-FCRM can be taken as two special cases of the proposed algorithm. Several experimental results are presented to demonstrate its advantage over FCRM in both noise insensitivity and robustness capability.
This paper presents a three-step framework to remove the highlight exists on objects in certain conditions. Unlike traditional HDR (High Dynamic Range) technology requires multiple registrated image sequence; our meth...
详细信息
This paper presents a three-step framework to remove the highlight exists on objects in certain conditions. Unlike traditional HDR (High Dynamic Range) technology requires multiple registrated image sequence; our method needs only two arbitrary images. SURF (Speeded Up Robust Features) matching algorithm is first applied to find corresponding point pairs between images; homography is then found by perspective transformation theory; minimum gray selection is used at last to eliminated the highlight and fuse related regions. Experimental results are given to demonstrate the performance of our method.
Single nucleotide polymorphisms (SNPs) are the most common form of genetic variant in humans, which can be generally classified into disease related mutations and common ones. It has been generally accepted that SNPs ...
详细信息
Single nucleotide polymorphisms (SNPs) are the most common form of genetic variant in humans, which can be generally classified into disease related mutations and common ones. It has been generally accepted that SNPs caused amino acid substitutions are of particular interest as candidates for affecting susceptibility to complex diseases, such as cancer, which is a serious public issue affecting millions of people worldwide each year. In this study, we have developed an automated and robust method to distinguish cancer-related mutations from common polymorphisms from amino acid sequence, which has a significant meaning for the cancer diagnosis, prognosis and treatment. Multiple different sequential features are extracted and the most important features are finally selected for constructing the prediction model. Experimental results show that an overall 81.07% success rate has been obtained, indicating the proposed method is very promising in the clinical cancer research studies.
This paper presents a novel facial localization method for 3D face in the presence of facial pose and expression variation. An idea of using Multi-level Partition of Unity (MPU) Implicits in a hierarchical way is prop...
详细信息
This paper presents a novel facial localization method for 3D face in the presence of facial pose and expression variation. An idea of using Multi-level Partition of Unity (MPU) Implicits in a hierarchical way is proposed for reconstruction of face surface. Based on the analysis of curvature features, nose and eyeholes regions can be detected on lower level reconstructed face surface uniquely. Experimental results show that this method is invariant to pose, holes, noise and expression. The overall performance of 99.18% is achieved.
Human facial shapes undergo significant variations from infancy to teenage, while they change limitedly as people age into adulthood. In this paper, facial shapes of males across ages using Euclidean distances are stu...
详细信息
Human facial shapes undergo significant variations from infancy to teenage, while they change limitedly as people age into adulthood. In this paper, facial shapes of males across ages using Euclidean distances are studied and a D-A standard (Distance related to Ages) is proposed for this purpose. Further, we determine the distance features that significantly contribute to the aging face recognition. Experimental results using the MORPH database containing age separated facial images are encouraging and illustrate Euclidean distance features improve the performance of face recognition across ages.
暂无评论