摘要:This paper presents a new scheme of fractal coding in the frequency domain. Compared with fractal coding in the spatial domain,the proposed scheme has two features: segmented luminance transform, simplification and combination of geometrical transform. Theory analysis and experiment show that this scheme is superior in encoding time and decoder contractility. Meanwhile, it shares alittle higher compression ratio and higher image fidelity.
摘要:The conventional iteration algorithm for generareing fractal images is in the complex plane.In this paper we define a fractal quasi-hologram and present a rendering method based on the hypercomplex number system for generating fractal quasi-holograms in more than four dimensions. The results indicate that the Mandelbrot and Julia sets are fractal and symmetrical in these higher dimensions. The fractal quasi-holograms can also be used in laser anti-counterfeiting field.
关键词:fractal;Quasi-hologram;Hypercomplex number system;Image synthesis
摘要:Image-Based technology have overwhelming advantage in independence of scene plexity.In this paper a new triangle feature based method is given, which depends on local information and produce realistic image. Compared with line feature based method,this algorithm obtains faster speed and will not twist the final result.
摘要:In this paper,one kind of segmentation methods of manmade object based on edge detection is introduced,which applies fractal,morphology and others comprehensively.This algorithm can keep more details than conventional methods while it can eliminate the edge in natural scenes background effectively.
摘要:A filter based on orthonormal wavelet bases is designed to do wavelet transform with images.Different scale image edges can be detected form different scale image wavelet transform. The different scale image edges are integrated to form high accurate image edges with only a pixel width.The detctor can also be used to determainate the singularity exponent of edges so as to identfy different types of edges.
摘要:Generalized Fuzzy Operator (GFO) is a new efficient algorithm for edge-detection of an image. By using of GFO, different display effects of an image can be obtained with different preprocessing methods.This paper gives an example of the relievo display of an image. A pre-processing method is introduced. When the parameters of this method are adjusted, the display is changed with them.
摘要:The analyses of images from remote sensing system are usually limited by the conditions such as time, equipment,fund and environment. In order to solve these problems, we determine the applicable conditions by means of statistics, set up the transfer model and simulated calculation formulas, explore and realize digital emulation for the remote-sensing image, by regarding the system as a“black box”.
摘要:The wavelet packet transform is used to detect the gray variation on textured images, then the multiresolution and multichannel texture features are extracted from every wavelet packets by the Envelop algorithm. These texture images with the same size are classified by the subspace classifier which can overcome the correlation between the feature images . The experiment indicates that the proposed approach provides a good results for textured image segmentation.
摘要:Image reconstruction is one of the most exciting borderline science in the last few decades. Many applications have become known such as in diagnostic radiology, nuclear medicine, industry checking and etc. We know that fan-beam projection is a widely used scheme in these applications. One disadvantage of fan-beam projection is the truncation problem in some case
which results in ring artifacts in the reconstruction. To resolve this problem, multi focal fan-beam projection is proposed in recent years to avoid the truncation problem and at the same time in the fan-beam projection' s good sensitivity especially for imaging of small organs is maintained. But the famous filering back projection method which is quick and efficient, can not be extended to such projection. In this thesis we advance a new method based on expanding fourier series and regularization technique. The method is as simple as the famous filtering back projection method and its simplified form can be applied to parallel-beam and fan-Beam cases. In implementing it, FFT is used to reduce the time of computation. Through computer simulation, it is found that the new method is noise-insensitivity and can produce accurate reconstruction almost without any visible artifacts when intensifying projection.
摘要:In this paper, an algorithm based on mathematics for improving the efficiency of WBS coding and
block coding of binary image is presented. The basic principle is to eliminate the black straight line sectors (vertical or horizontal) by preprocessing them before encoding, the compression ratio is then increased.
关键词:Binary image;compression ratio;Two-dimensional standard trasformation
摘要:We present a novel 2D metamophosis technique,which uses a distance map to link two shapes.There are three steps.Firstly,the position of a shape is calculated.Secondly,the shape is blended with a distance map.Finally an iterative process isused to linearly interpolate the positions and orientations,and non-linearly interpolate the shapes using morphology.The proposed 2D metamorphosis technique is used for slice interpolation with MRI stack images.The result of shape blending and 3D surface displayingis given.
摘要:The geometry model of binocular perception are researched in this paper by utilizing double video camera model of normal needle apertune. It discusses visual reconstruction of stereographics vision image space on detail and analyzes the factors influencing stereographics vision by experiments finally.
关键词:Binocular perception;Double video camera model;Stereo vision factor;virtual reality
摘要:An algorithm using Hopfield neural network to recognize intersection lines is presented in this paper.This algorithm utilizes a new concept which is much different from the conventional processing methods of intersection lines recognition. In this algorithm, an image containing intersection lines is mapped to a Hopfield neural networks.Every pixel in the image is mapped to a neuron, and the lines between neurons are determined by the relations of the represented pixels such as the distance, potentials difference and slope. During the Hopfield neural nets evaluation, the neurons states are modified till convergence. Along the settlement of the neurons' states, the mapped intersection lines is recognized. During experimental simulations, this algorithm really showed fine performance.
摘要:Multimedia synchronization is a significant requirement for distributed multimedia applications.In this paper,we augment the OCPN model,and present a new multimedia synchronization model XOCPN based on timed Petri nets. Using this model we can give the formal specification for multimedia object compositions with time-constrained relations and deal
with flexible user interaction. Algorithms for asynchronous user interactions are also presented, and explained by an example.
关键词:Timed Petri nets;Multimedia synchronization;Asynchronization user interaction
摘要:The problem of multimedia synchronization must be resolved on the multimedia presentation,and intra-media synchronization is its basis. In this paper, we propose an approach for allocating buffer N and give their time relationship figures, mathematics description and induction, with restricted blocking policy and blocking policy based on linked distributed networks and its QoS parameters. We analyze and demonstrate its reliability and correctness.