In this paper, we apply a multiple regression method based on canonical correlation analysis (CCA) to face data modelling. CCA is a factor analysis method which exploits the correlation between two high dimensional si...
详细信息
In this paper, we apply a multiple regression method based on canonical correlation analysis (CCA) to face data modelling. CCA is a factor analysis method which exploits the correlation between two high dimensional signals. We first use CCA to perform 3D face reconstruction and in a separate application we predict near-infrared (NIR) face texture. In both cases, the input data are color (RGB) face images. Experiments show, that due to the correlation between input and output signal, only a small number of canonical factors are needed to describe the functional relation of RGB images to the respective output (NIR images and 3D depth maps) with reasonable accuracy
The functional network was introduced by ***, which extended the neural network. Not only can it solve the problems solved, but also it can formulate the ones that cannot be solved by traditional network. This paper a...
详细信息
The functional network was introduced by ***, which extended the neural network. Not only can it solve the problems solved, but also it can formulate the ones that cannot be solved by traditional network. This paper applies functional network to approximate the multidimension function under the ridgelet theory. The method performs more stable and faster than the traditional neural network. The numerical examples demonstrate the performance.
An algorithm for traversing voxels along a 3D line using only integer arithmetic is presented. Two kinds of subdivision, namely non-unit space subdivision and non-square space subdivision, could be used. A 2D algorith...
详细信息
An algorithm for traversing voxels along a 3D line using only integer arithmetic is presented. Two kinds of subdivision, namely non-unit space subdivision and non-square space subdivision, could be used. A 2D algorithm is first developed to find out the pixels a line passes through. Then based on it, an algorithm for traversing voxels along a 3D line in uniformly divided voxel space is proposed. Up to 3 voxels may be processed in one step and no accumulated error is made. The discriminator used in this algorithm is well refined, so that few calculations are required and the algorithm is simple and compact in form. Its comparison with existing voxel traversal algorithms shows that the new algorithm is not only free from accumulated error but also is the fastest.
In this paper active feature models are proposed. They utilize local texture features and a statistical shape model for the reliable localization of landmarks in images. They are related to active appearance models, b...
详细信息
ISBN:
(纸本)0769525210
In this paper active feature models are proposed. They utilize local texture features and a statistical shape model for the reliable localization of landmarks in images. They are related to active appearance models, but instead of modelling the entire texture of an object they represent image texture by means of local descriptors. The approach has advantages with complex image data like anatomical structures that exhibit high texture variation with limited relevance for the recognition of the object location. Experimental results and the comparison to AAMs on different data sets indicate that active feature models can improve search speed and result accuracy, considerably
This paper presents a method for the fully automatic surveying of cutaneous hemangiomas by means of a hemangioma segmentation and a ruler visible in the images. The algorithm computes the spatial resolution of an imag...
详细信息
This paper presents a method for the fully automatic surveying of cutaneous hemangiomas by means of a hemangioma segmentation and a ruler visible in the images. The algorithm computes the spatial resolution of an image. Hemangioma segmentation is accomplished by a single-layer perceptron classification by means of pixel color features. The algorithm was evaluated on a set of 120 images. It achieves satisfactory results on images with clearly visible, saturated hemangiomas
Autonomous model building is a crucial trend in model based methods like AAMs. This paper introduces an approach that deals with non-linearities by detecting distinct sub-parts in the data. Sub-models each representin...
详细信息
ISBN:
(纸本)1901725294
Autonomous model building is a crucial trend in model based methods like AAMs. This paper introduces an approach that deals with non-linearities by detecting distinct sub-parts in the data. Sub-models each representing an individual sub-part are derived from a minimum description length criterion. Thereby the resulting clique of models is more compact and obtains a better generalization behavior than a single model. The proposed AAM clique generation deals with non-linearities in the data in a generic information theoretic manner reducing the necessity of user interaction during training.
This paper proposes a method for robustly matching active appearance models (AAMs) on images with gross disturbances (outliers). The method consists of two steps. First, an initial residual is calculated by comparing ...
详细信息
ISBN:
(数字)9783540264316
ISBN:
(纸本)3540250522
This paper proposes a method for robustly matching active appearance models (AAMs) on images with gross disturbances (outliers). The method consists of two steps. First, an initial residual is calculated by comparing model and image appearance, and modes of the residual are analyzed. Second, all possible mode combinations are tested by evaluating an objective function. The objective function allows the selection of an outlier-free mode combination. Experiments demonstrate the ability of the robust matching method to successfully cope with outliers - compared to standard AAM matching, no degeneration of the model during matching occurs.
We present two approaches for automatically segmenting the spinal cord/canal from native CT images of the thorax region containing the spine. Different strategies are included to handle images where only part of the s...
详细信息
Modifying some knots of a B-spline or NURBS surface of order k×h to generate a family of B-spline or NURBS surface is presented. We show that the envelope of the family is a B-spline or NURBS surface defined by t...
详细信息
Modifying some knots of a B-spline or NURBS surface of order k×h to generate a family of B-spline or NURBS surface is presented. We show that the envelope of the family is a B-spline or NURBS surface defined by the same control points, and its order is (k-a)×(h-b), where a, b are the multiplicity of the modified knots. Moreover, any order partial derivatives of the B-spline surface and the family of B-spline surfaces differ only in a multiplier. The presented results can be served as a theoretical guidance for surface modeling and shape modifications in computer aided design today.
Active shape models are powerful and widely used tool to interpret complex image data. By building models of shape variation they enable search algorithms to use a priori knowledge in an efficient and gainful way. How...
详细信息
Active shape models are powerful and widely used tool to interpret complex image data. By building models of shape variation they enable search algorithms to use a priori knowledge in an efficient and gainful way. However, due to the linearity of PCA, non-linearities like rotations or independently moving sub-parts in the data can deteriorate the resulting model considerably. Although non-linear extensions of active shape models have been proposed and application specific solutions have been used, they still need a certain amount of user interaction during model building. In this paper the task of building/choosing optimal models is tackled in a more generic information theoretic fashion. In particular, we propose an algorithm based on the minimum description length principle to find an optimal subdivision of the data into sub-parts, each adequate for linear modeling. This results in an overall more compact model configuration. Which in turn leads to a better model in terms of modes of variations. The proposed method is evaluated on synthetic data, medical images and hand contours.
暂无评论