this paper describes different approaches of realtime GMM (Gaussian Mixture Method) background subtraction algorithm using video sequences for hand image segmentation. In each captured image, the segmentation takes pl...
详细信息
ISBN:
(纸本)0769526861
this paper describes different approaches of realtime GMM (Gaussian Mixture Method) background subtraction algorithm using video sequences for hand image segmentation. In each captured image, the segmentation takes place where pixels belonging to the hands are separated from the background based on background extraction and skin-color segmentation. A time-adaptive mixture of Gaussians is used to model the distribution of each pixel color value. For an input image, every new pixel value is checked, deciding if it matches with one of the existing Gaussians based on the distance from the mean in terms of the standard deviation. the best matching distribution parameters are updated and its weight is increased. It is assumed that the values of the background pixels have low variance and large weight. these matched pixels, considered as foreground, are compared based on skin color thresholds. the hands position and other attributes are tracked by frame. that enables us to distinguish the hand movement from the background and other objects in movement, as well as to extract the information from the movement for dynamic hand gesture recognition.
the choice of a color model is of great importance for many computer vision algorithms. However, there are many color models available;the inherent difficulty is how to automatically select a single color model or, al...
详细信息
ISBN:
(纸本)9780769529967
the choice of a color model is of great importance for many computer vision algorithms. However, there are many color models available;the inherent difficulty is how to automatically select a single color model or, alternatively, a subset of features from several color models producing the best result for a particular task. To achieve proper colors components selection, in this paper, it was proposed the use of wrapper method, a data mining approach, to obtain repeatability and distinctiveness in segmentation process. the result was compared with neural network method and yields good feature discrimination. the method was verified experimentally with 108 images from Amsterdam Library of Objects images (ALOI) and 10 aerial images with different photometric conditions. Furthermore, it has shown that the color model selection scheme provides a proper balance between color invariance (repeatability) and discriminative power (distinctiveness).
this paper exposes an algorithm which leads to a fuzzy segmentation. this algorithm performs, as in the watershed method, a progressive flood of the gradient image from pixels of lowest gradients. It uses a new distan...
详细信息
ISBN:
(纸本)0769513301
this paper exposes an algorithm which leads to a fuzzy segmentation. this algorithm performs, as in the watershed method, a progressive flood of the gradient image from pixels of lowest gradients. It uses a new distance, called topographic distance. Any local minimum of the gradient norm image constitutes a seed for the region g-rowing, avoiding the use of a marker image. these seeds constitute the cores of the initial fuzzy regions. then the sites are gradually agglomerated to the region, while their membership degrees to the region decrease, according to the distance to the core and to the gradient norms, by the way of the topographic distance. the numerous fuzzy regions are then merged and the membership degrees of pixels to final regions are computed. Applications concern crisp segmentation of colour or gray scale images and pattern recognition from fuzzy regions.
images rendered using global illumination algorithms are considered amongst the most realistic in 3D computergraphics. However this high fidelity comes at a significant computational expense. A major part of this cos...
详细信息
ISBN:
(纸本)0769523897
images rendered using global illumination algorithms are considered amongst the most realistic in 3D computergraphics. However this high fidelity comes at a significant computational expense. A major part of this cost arises from the sampling required to eliminate aliasing errors. these errors occur due to the discrete sampling of continuous geometry space inherent to these techniques. In this paper we present a fast analytic method for predicting in advance where antialiasing needs to be computed. this prediction is based on a rapid visualisation of the scene using a GPU, which is used to drive a selective renderer We are able to significantly reduce the overall number of anitialiasing rays traced, producing an imagethat is perceptually indistinguishable from the high quality image at a much reduced computational cost.
Texture information is useful for image data browsing and retrieval. the goal of this paper is to present a texture classification system for remote sensing images addressed to the administration of great collections ...
详细信息
ISBN:
(纸本)076951846X
Texture information is useful for image data browsing and retrieval. the goal of this paper is to present a texture classification system for remote sensing images addressed to the administration of great collections of those images. the proposed classifier is a hybrid system composed by an unsupervised neural network and a supervised one. Starting from a small portion of the image (pattern) the system should recognize the most similar class to a pattern in a database as well as to identify the images that contain similar patterns. the texture feature vectors used to characterize the patterns are obtained from the images processed by a bank of Gabor Filters. Some experimental results using textures of the Brodatz album, multi-spectral and radar images are presented here.
Texture segmentation is a basic task in the process of satellite image analysis. However, texture characterization is not an easy task specially when more than one wavelength band need to be considered. Several diffic...
详细信息
Digital image interpolation techniques are frequently used to enlarge pictures, i.e. zooming in, and they are upon increasingly demand for developing product applications using digital still cameras. One common diffic...
详细信息
ISBN:
(纸本)076951846X
Digital image interpolation techniques are frequently used to enlarge pictures, i.e. zooming in, and they are upon increasingly demand for developing product applications using digital still cameras. One common difficulty with conventional interpolation techniques is that of preserving details, i.e. edges, and at the same time smoothing the data for not introducing spurious artifacts. A definitive solution to this is still an open issue, although there are working methods in the market, see e.g. Parker et. al. [6], Sakamoto et. al. [8] for recent surveys. In this paper we propose a locally adaptive edge-preserving algorithm for image interpolation, which deals withthis problem, and different than other methods shows how to compute local thresholds preserving edges and not destroying smoothness at the same time.
We present an algorithm for polygonizing closed implicit surfaces, which produces meshes adapted to the local curvature of the surface. Our method is similar to, but NOT based on, Marching Triangles, in that we start ...
详细信息
ISBN:
(纸本)0769522270
We present an algorithm for polygonizing closed implicit surfaces, which produces meshes adapted to the local curvature of the surface. Our method is similar to, but NOT based on, Marching Triangles, in that we start from a point on the surface and develop a mesh from that point using a surface-tracking approach. In a marked departure from previous approaches, our meshes approximate the surface through heuristics relying on curvature. Furthermore, our method works completely on-the-fly, resolving cracks as it proceeds, without the need for any post-remeshing step to correct failures. We have tested the algorithm withthree different representations of implicit surfaces, Variational, analythical and MP U, using non-trivial data sets, yielding results that illustrate the flexibility and scalability of our technique. Performance comparisons with variants of Marching Cubes show that our approach is capable of good accuracy and meshing quality without sacrificing computing resources.
Nowadays, a personal music collection may comprise thousands of MP3 files. Visualization can help the user to gain an overview and to find similar songs inside so large a set. We describe a method to create icons from...
详细信息
ISBN:
(纸本)0769526861
Nowadays, a personal music collection may comprise thousands of MP3 files. Visualization can help the user to gain an overview and to find similar songs inside so large a set. We describe a method to create icons from audio files in such a way that songs which the user considers similar receive similar icons. this allows visual data mining in standard directory listings of window-based operating systems. the icons consist of bloom-like shapes, whose form and color depend on eight parameters. these parameters are controlled through a neural net, the input of which are audio features that are extracted algorithmically from the MP3 files. To adapt the system to the user's perception and interests, the neural net is initially trained with a small set of songs and icons. User studies done on the system demonstrate a strong perceptual relation between music and icons.
We present a model for geometric modeling and rendering of individual feathers for computergraphics purposes. the model represents the feather structure with a collection of parameterized Bezier curves. this parametr...
详细信息
ISBN:
(纸本)076951846X
We present a model for geometric modeling and rendering of individual feathers for computergraphics purposes. the model represents the feather structure with a collection of parameterized Bezier curves. this parametrization allows easy generation of the existent types of feather structures with biologically-motivated parameters. Once the feather structure is defined, the feathers can be rendered either in a non-photorealistic rendering style or using texture mapping to achieve a more natural look.
暂无评论