Algorithm for the Removing Uniformed Motion Blur

Motion deblurring is one of the basic problems inthe field of image processing. This paper summarizes the mathematical basis of the previous work and presents a deblurringmethod that can improve the estimation of the motion blurkernel and obtain a better result than the traditional methods.Experiments show the motion blur kernel loses some important and useful properties during the estimation of the kernel which may cause a bad estimation and increase the ringingartifacts. Considering that the kernel is provided by the motion of the imaging sensor during the exposure and that the kernel shows the trace of the motion, this paper ensures the physical meaning of the kernel such as the continuity and the center of thekernel during the iterative process. By adding a post process to the estimation of the motion blur kernel, we remove some discrete points and make use of the centralizationof the kernel in order to accurate the estimation. The experiment shows the existence of the post process improves the effect of the estimation of the kernel and provides a better result with the clear edges.     

Combination of wavelet and SIFT features for image classification using trained Gaussian mixture model

This paper presents an effective combination of Wavelet-based features and SIFT features. For the combined feature patches extracted from images we then adopt the PCA transformation to reduce the dimensionality of their feature vectors. And the reduced vectors are used to train Gaussian Mixture Models （GMMs） in which the mixture weights and Gaussian parameters are updated iteratively. We performed the method on Caltech datasets and compared the results with several other methods. It shown that the combination of salient feature vectors and GMM gives a much better improvement in image classification.     

Super-Resolution image Reconstruction Based on Iteration and Wavelet Transform

In this paper, we provide a super-resolution image reconstruction algorithm based on wavelet transform. Wavelet transform can separate high frequency and low frequency information of image. The more high frequency information can be obtained by using wavelet transform and the technique of image fusion. Meanwhile, reconstructed super-resolution image is produced by the iterative method. In iteration process, noise of image can be suppressed by applying method of wavelet threshold de-noising. The experiment results show that the algorithm can overcome the disadvantage of the classical interpolation method and effectively improve the resolution and PSNR of the image.     

Multi-agent based evolutional algorithm in medicine image segmentation

When we analyses medicine image, many research objects （such as human brain and heart, et al） have no obvious brim, it is nature texture image and hasn＇t clear boundary between different organizations. An evolutional algorithm of medicine image segmentation based on multi-agent system is proposed in this paper. The agent is designed to be distributed calculation entities of the four entities of the calculation of the distribution in the algorithm, it presence at the two-dimensional grid in the medicine image, using the priori knowledge to guide its evolution. Through the human brain MR image segmentation experiment, compare to maximum likelihood segmentation and Conjugate Gradient Square segmentation, our method is more suitable for clinical.     

An application to detect the edge and texture of the flower by canny algorithm

For the edge and texture characteristics of the flower, a new improved canny algorithm is proposed. In the method, canny algorithm detects the edge and the texture of the flower perfectly, the image noise is eliminated simultaneously. The method is applied to all bloom stages. The growing status of the flower could be better analyzed.     

Efficient dark channel based image dehazing using quadtrees

Using dark channel prior—a kind of statistics of the haze-free outdoor images—to remove haze from a single image input is simple and effective. However, due to the use of soft matting algorithm, the method suffers from massive consumption of both memory and time, which largely limits its scalability for large images. In this paper, we present a hierarchical approach to accelerate dark channel based image dehazing. The core of our approach is a novel, efficient scheme for solving the soft matting problem involved in image dehazing, using adaptively subdivided quadtrees built in image space. Acceleration is achieved by transforming the problem of solving a N-variable linear system required in soft matting, to a problem of solving a much smaller m-variable linear system, where N is the number of pixels and m is the number of the corners in the quadtree. Our approach significantly reduces both space and time cost while still maintains visual fidelity, and largely extends the practicability of dark channel based image dehazing to handle large images.     

Local thresholding with adaptive window shrinkage in the contourlet domain for image denoising

Threshold selection is a challenging job for the image denoising in the contourlet domain. In this paper, a new local threshold with adaptive window shrinkage is proposed. According to the anisotropic energy clusters in contourlet subbands, local adaptive elliptic windows are introduced to determine the neighboring coefficients with strong dependencies for each coefficient. Utilizing the maximum likelihood estimator within the adaptive window, the signal variance is estimated from the noisy neighboring coefficients. Based on the signal variance estimation, the new threshold is obtained in the Bayesian framework. Since it makes full use of the captured directional information of images, the threshold extends to the anisotropic spatial adaptability and behaves reliably. Simulation experiments show that the new method exhibits better performance than other outstanding wavelet and contourlet denoising schemes obviously, both in the PSNR value and the visual appearance.     

An improved clustering analyzing algorithm for image index

This paper provides a survey of current clustering analysis in image indexing at first, then the problems are solved and the new directions in the future in this domain are pointed out. In the end, it proposes a new clustering algorithm with the combination of k-means clustering algorithm, genetic algorithm and chaotic optimization, a new idea of dimensionality reduction.     

Content-based facial image retrieval using common vector

Abstract： Facial image retrieval is an essential application of content-based image retrieval. Based on the analysis of the practical application background, this paper proposes a new facial image retrieval scheme. In this scheme, the input query image is firstly transformed by four different methods to generate virtual samples and enlarge the training set. Moreover common vector method is applied to span the feature space for the training set whose images just belong to one class. To prove the feasibility of the scheme, a series of experiments are performed on the ORL face database.     

Medical Image Retrieval Based on Multi-Layer Resampling Template

Medical image application in clinical diagnosis and treatment is becoming more and more widely, How to use a large number of images in the image management system and it is a very important issue how to assist doctors to analyze and diagnose. This paper studies the medical image retrieval based on multi-layer resampling template under the thought of the wavelet decomposition, the image retrieval method consists of two retrieval process which is coarse and fine retrieval. Coarse retrieval process is the medical image retrieval process based on the image contour features. Fine retrieval process is the medical image retrieval process based on multi-layer resampling template, a multi-layer sampling operator is employed to extract image resampling images each layer, then these resampling images are retrieved step by step to finish the process from coarse to fine retrieval.     

Image decomposing for inpainting using compressed sensing in DCT domain

Inpainting images with occlusion or corruption is a challenging task. Most existing algorithms are pixel based, which construct a statistical model from image features. However, in these algorithms, the frequency component is not sufficiently addressed. In this paper, we propose a novel algorithm that utilizes compressed sensing （CS） in frequency domain to reconstruct corrupted images. In order to reconstruct image, we first decompose the image into two functions with different basic characteristics - structure component and textual component. We seek a sparse representation for the functions and use the DCT coefficients of this representation to generate an over-complete dictionary. Experimental results on real world datasets demonstrate the efficacy of our method in image inpainting. We compare our method with three state-of-the-art inpalnting algorithms and demonstrate its advantages in terms of both quantitative and qualitative aspects.     

On the ?2 stability of time varying linear and nonlinear discrete time MIMO systems

New conditions are derived for the l2-stability of time-varying linear and nonlinear discrete-time multiple-input multipleoutput （MIMO） systems, having a linear time time-invariant block with the transfer function F（z）, in negative feedback with a matrix of periodic/aperiodic gains A（k）, k = 0,1, 2,... and a vector of certain classes of non-monotone/monotone nonlinearities φp（-）, without restrictions on their slopes and also not requiring path-independence of their line integrals. The stability conditions, which are derived in the frequency domain, have the following features： i） They involve the positive definiteness of the real part （as evaluated on ｜z｜ = 1） of the product of Г （z） and a matrix multiplier function of z. ii） For periodic A（k）, one class of multiplier functions can be chosen so as to impose no constraint on the rate of variations A（k）, but for aperiodic A（k）, which allows a more general multiplier function, constraints are imposed on certain global averages of the generalized eigenvalues of （A（k ＋ 1）,A（k））, k = 1, 2 iii） They are distinct from and less restrictive than recent results in the literature.     

Context-awareness approach in game style application

By combining training content with Game AI, Game Style Training can provide a rich, interactive and engaging educational substance. It supports experiential learning-bydoing. And it can contribute to promote interest and motivation of trainees. In this paper, a Context-Awareness Approach （CAA） is put forward for earthquake rescue training by applying the concept of ambient intelligence. The CAA takes advantage of player profile to monitor player states, utilizes time sensor and location sensor to supervise environment changes. And then, a Reinforcement Learning model based on MAS is designed to update game AI system.     

Whitening of fractional order polynomial autoregressive Non-Gaussian impulsive stable colored noise

This paper proposes a new different spectral density from power spectrum density of second order processes, thus we can get a new concept of stable white noise based on covariation function and covariation spectrum density. And a new technique is suggested for the estimation of the parameters of polynomial autoregressive （PAR） nonlinear colored noise processes with stable white noise excitations. Simultaneously, a simple whitening method is employed for polynomial autoregressive nonlinear stable colored noise.     

Image decomposition on the basis of an inverse pyramid with 3-layer neural networks

The contemporary information technologies and Internet impose high requirements on the image compression efficiency. Great number of methods for information redundancy reduction had already been developed, which are based on the image processing in the spatial or spectrum domain. Other methods for image compression use some kinds of neural networks. In spite of their potentialities, the methods from the last group do not offer high compression efficiency. New adaptive method for image decomposition on the basis of an inverse pyramid with neural networks is presented in this paper. The processed image is divided in blocks and then each is compressed in the space of the hidden layers of 3-layer BPNNs, which build the so-called inverse difference pyramid. The results of the new method modeling are presented for sequence of static images in comparison with results for single images from the same group.     

Topology control for directed diffusion in wireless sensor networks

Due to severely constrained energy, storage capacity and computing power for nodes in wireless sensor network, the routing protocols have become a hot research topic. Directed diffusion is a communication paradigm for information dissemination in sensor networks based on data-centric routing. An energy efficient hybrid adaptive clustering for directed diffusion is presented in the paper. The main goal is to curb the interests flooding overhead by clustering. Hybrid of active and passive clustering creation is used to make tradeoff between maintenance cost and delay. The adaptive turning off the radio of redundant nodes and rotation of cluster heads is used to save energy. The ns-2 simulating results show that the protocol has good performances in energy consuming, delivery ratio and delay without leading to too much overhead comparing with existing directed diffusion protocols.     

Finite Element Analysis of Steam Turbine Diaphragm

Thediaphragmofsteam turbine isoneofthemainpartsofthepassageunitfor air flowing.For studying its working conditionand performance, the three-dimensionalfinite element model（FEM）with thesoftware ANSYS10.0 isestablished according to theactualworking parameters. With the model, the deformation resultsand stress field resultsare analyzed andthe maximum displacement and stress value are calculated, meanwhile,theircorrespondingpositions are found. It provides a good foundationfor solving the problem inproduction.     

Image geometry and radio metrics correction and optimization in image mosaic

In image guided navigation, due to radiometric and geometric distortions of images from photography, it can mislead the content of the image, image series are finally converted into the same coordinate system. Based on the analysis of correction algorithm, a much better one is presented, which is proved to be effective in the correction of radiometric image distortion. A theory using image local maximum modules, got by wavelet transform, as image feature, is promoted. Using image mosaic technique based on multi-resolution analysis, redundancies between images are eliminated. The whole image produced at last is a much bigger one without seam and distortion.     

DNA-chip-based dynamic broadcast encryption scheme with constant-size ciphertexts and decryption keys

In this paper we present DNA-DBE,a DNA-chip-based dynamic broadcast encryption scheme.In our scheme,new users can join dynamically without modification of other users’decryption keys.Either the ciphertext or the decryption key is of constant-size.Backward secrecy is achieved in DNA-DBE:if new users join the system dynamically,they will not be able to retrieve past data.The security of our scheme relies on hard biological problems,which are immune to attacks of new computing technologies in the future.There exists a special feature in DNA–based cryptosystems,i.e.the set of encryption keys and the set of decryption keys have a many-to-many relationship.The implementation of more complicated DNA cryptosystems taking advantage of this special feature has been previously left as an open problem.Our DNA-DBE system is a solution to this open problem,which is also the first exploration of DNA based group-oriented encryption system.