Search

Face aging aims at synthesizing one΄s face at different ages which interests many researchers in fields of cartoon animation, age estimation, face recognition, etc. However, modelling aging process is still challenging due to lack of robust features and a reasonable training set. In this paper we propose a novel automatic face aging approach through Maximum A Posteriori (MAP), a face is firstly warped by global geometric transformation, and then elder skin is locally synthesized by sparse coding. Our contribution includes an aging model both for children growth and adults aging, and high-level features by sparse representation aiming at a small training set while not downgrading the quality of synthesis. Moreover, the newly proposed features simplify the algorithm and lead to a fast implementation. Experiments show the proposed approach outperforms the existing methods. [ABSTRACT FROM AUTHOR]

Human action recognition is an important problem in computer vision. Most existing techniques use all the video frames for action representation, which leads to high computational cost. Different from these techniques, we present a novel action recognition approach by describing the action with a few frames of representative poses, namely kPose. Firstly, a set of pose templates corresponding to different pose classes are learned based on a newly proposed Pose-Weighted Distribution Model (PWDM). Then, a local set of kPoses describing an action are extracted by clustering the poses belonging to the action. Thirdly, a further kPose selection is carried out to remove the redundant poses among the different local sets, which leads to a global set of kPoses with the least redundancy. Finally, a sequence of kPoses is obtained to describe the action by searching the nearest kPose in the global set. And the proposed action classification is carried out by comparing the obtained pose sequence with each local set of kPose. The experimental results validate the proposed method by remarkable recognition accuracy. [ABSTRACT FROM AUTHOR]

Visual attention is an important issue in image and video analysis and keeps being an open problem in the computer vision field. Motivated by the famous Helmholtz principle, a new approach of visual attention analysis is proposed in this paper based on the low level feature statistics of natural images and the Bayesian framework. Firstly, two priors, i.e., Surrounding Feature Prior (SFP) and Single Feature Probability Distribution (SFPD) are learned and integrated by a Bayesian framework to compute the chance of happening (CoH) of each pixel in an image. Then another prior, i.e., Center Bias Prior (CBP), is learned and applied to the CoH to compute the saliency map of the image. The experimental results demonstrate that the proposed approach is both effective and efficient by providing more accurate and quick visual attention location. We make three major contributions in this paper: (1) A set of simple but powerful priors, SFP, SFPD and CBP, are presented in an intuitive way; (2) A computational model of CoH based on Bayesian framework is given to integrate SFP and SFPD together; (3) A computationally plausible way to obtain the saliency map of natural images based on CoH and CBP. [ABSTRACT FROM AUTHOR]

Image retrieval on large-scale datasets is challenging. Current indexing schemes, such as k-d tree, suffer from the ˵curse of dimensionality″. In addition, there is no principled approach to integrate various features that measure multiple views of images, such as color histogram and edge directional histogram. We propose a novel retrieval system that tackles these two problems simultaneously. First, we use random projection trees to index data whose complexity only depends on the low intrinsic dimension of a dataset. Second, we apply a probabilistic multiview embedding algorithm to unify different features. Experiments on MSRA large-scale dataset demonstrate the efficiency and effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

Implicit surface reconstruction has been a challenging work for decades. In this paper, motivated by the concept of classic physical polar field model and off-set points strategy, we present a new approach, called Field Fitting. In this approach, we express a 3D surface as an equipotential surface of scalar polar field which is produced by a number of paired field generating primitives, then a surface reconstruction process is cast as a primitives localization process, and finally, we solve this problem with a greedy method. Experimental results demonstrate that the proposed method outperforms the previous by providing better surface reconstruction results. [ABSTRACT FROM AUTHOR]

We present a novel algorithm for speech emotion classification. In contrast to previous methods, we additionally consider the relations between simple features by incorporating covariance matrices as the new feature descriptors. Since non-singular covariance matrices do not lie on a linear space, we endow the space with an affine invariance metric and render it into a Riemannian manifold. After that we use the tangent space to approximate the manifold. Classification is performed in the tangent space and a generalized principal component analysis is presented. We test the algorithm on speech emotion classification and the experiment results show an improvement at around 13%(+3% with PCA) in recognition accuracy. Based on that we are able to train one simple model to accurately differentiate the emotions from both genders. [ABSTRACT FROM AUTHOR]

The automatic recognition of facial expression presents a significant challenge to the pattern analysis and man-machine interaction research community. In this paper, a novel system is proposed to recognize human facial expressions based on the expression sketch. Firstly, facial expression sketch is extracted by an GPU-based real-time edge detection and sharpening algorithm from original gray image. Then, a statistical method, which is called Eigenexpress, is introduced to obtain the expression feature vectors for sketches. Finally, Modified Hausdorff distance(MHD) was used to perform the expression classification. In contrast to performing feature vector extraction from the gray image directly, the sketch based expression recognition reduces the feature vector's dimension first, which leads to a concise representation of the facial expression. Experiment shows our method is appreciable and convincible. [ABSTRACT FROM AUTHOR]

There are certain correlation between two persons' emotional states in communication, but none of previous work has focused on it. In this paper, a novel conversation database in Chinese was collected and an emotion interaction matrix was proposed to embody the discourse information in conversation. Based on discourse information, an enhanced speech emotion recognition system was presented to improve the recognition accuracy. Some modifications were performed on traditional KNN classification, which could reduce the interruption of noise. Experiment result shows that our system makes 3% - 5% relative improvement compared with the traditional method. [ABSTRACT FROM AUTHOR]

In this paper, a novel technique is presented to incorporate vector field based feature extraction schemes into painterly rendering. This approach takes a raster photograph as input and automatically creates a hand-painting style picture. Via techniques formerly used in image segmentation, a vector field representation are generated, identifying color and texture variations at each pixel location, and a series of brush strokes are created with sizes and alignments controlled by the vector field and color matched from the original picture. Moreover, different scale parameters could be utilized to produce several vector fields depicting images features of the original photograph from rough outline to detail. The final output could be rendered first by brushstrokes in the coarsest scale and refined progressively. Unlike conventional techniques that used taking account only of local color gradients, this approach employs multi-scale feature extraction scheme to guide stroke generation with image structure on larger scale. [ABSTRACT FROM AUTHOR]

We present a learning-based 3D facial expression mapping technique that preserves facial expression details and works in real time for even high resolution meshes. Our approach is inspired by a previously developed technique called deformation transfer [1]. The deformation transfer technique preserves facial expression details but its computational overhead makes it not suitable for real time applications. To accelerate computation, we use a piecewise linear function to represent the mapping from the direct motion (the difference between the expression face and the neutral face) to the motion obtained by the deformation transfer method. This piecewise linear function is learned offline from a small set of training data. The online computation is thus reduced to the evaluation of the piecewise linear functions which is significantly faster. As a result, we are able to perform real time expression mapping for even high resolution 3D meshes. [ABSTRACT FROM AUTHOR]