How to select the hidden area for watermarking is an important issue in the design of watermarking algorithm. In this paper an adaptive DCT watermarking algorithm is proposed, in which the hidden area is selected according to human visual system model and local characteristics of image. Thus robustness and imperceptibility could be well balanced. Experimental results demonstrate that this algorithm is effective, and invisible watermark is robust against many kinds of image processing such as additive white Gaussian noise, JPEG compression, cropping, median filtering and resizing.

In this paper, a new algorithm of image watermark embedding and detecting based on DCT transform is presented. This method is robust to common signal and geometric distortions such as image cropping, and the embedded watermark can be detected without resorting to the uncorrupted original image. Experimental results support these claims.

An algorithm that can synthesize multi-pose face images from a single view or small samples is proposed in this paper. First, the human face is represented with a dominant point set. Then, the variance of the dominant point set between different poses is fitted based on a least square fitting with a polynomial function and a global morphing field is formed. Finally, multi-pose face images are synthesized by image warping from a single view based on the global morphing field. An experiment of multi-pose face recognition is done based on the single view and synthesized multi-pose images. The results show that its performance is by far superior to those of the traditional methods. Thus, they prove the effectiveness of the algorithm in this paper.

By combining the model-based analysis method for dynamic systems with multi-scale signal transformation method based on statistical characteristics, this paper proposed a group of multiscale dynamic models and multiscale error models based on the dynamic systems which consist of a state model given at the finest scale and multisensor having different sample rates at different scales. A new multiscale recursive fusion estimation algorithm is put forward by use of the recursive relationship between scales. At the finest scale the optimal fusion estimates are obtained on the basis of global information. Finally the validity of the method is proved and the effectiveness of the new algorithm is illustrated by use of an example.

A Tree-structured Growing Self-Organizing Maps (TGSOM) is presented as an extended version of the Self-Organizing Feature Maps (SOFM), which has significant advantages for data mining applications. The TGSOM algorithm is presented in detail and the effect of a spread factor, which can be used to measure and control the spread of the TGSOM, is investigated. The spread factor is independent of the dimensionality of the data and as such can be used as a controlling measure for generating maps with different dimensionality, which can then be compared and analyzed with better accuracy. The spread factor is also presented as a method of achieving hierarchical clustering of a data set with the TGSOM. Such hierarchical clustering allows the data analyst to identify significant and interesting clusters at a higher level of the hierarchy, and as such continue with finer clustering of only the interesting clusters.

A new invariants classification feature based on cyclic temporary cumulant is proposed for classification of MPSK signals. The new feature only uses the symbol rate information, which is resistant to shaping pulse and invariant with respect to the time-translation, local carrier phase offset and amplitude scale of the baseband MPSK signals, and can suppress stationary additive noise. The signal selectivity of cyclic temporary cumulants makes it possible for the proposed feature to be applied to modulation classification of multi-signals. Theoretial analysis and extensive computer simulations show the efficiency of the proposed classification algorithm.

Based on overlapping individual filter character, IIR general parameter filters are used as units of filterbank, whose amplitude and delay characters are all considered. In designing, let filter frequency character in passband region meet the demand, then optimizing the integrated frequency character in transition region with some anti-aliasing or anti-crosstalk measures used, the resulting filter bank can be nearly perfect reconstruction in certain precision. This is a new theory and method of design for filter banks.

The higher-order statistics has found applications in the wide area of signal processing problem. However in some practical applications, only a sample of the objective process can be obtained. Thus the statistical average must be replaced by time average. The precondition to do this is the ergodicity of the processes. The ergodicity of the higher-order statistics of the harmonic processes is analyzed in this paper. Further more, the applications of the ergodicity in the phase coupling recognition and the array processing problems using higher-order statistics are proposed in this paper. Simulation examples verified those results.

ESPRIT can estimate one direction of signals utilizing a single shift-invariance property of the sensor array. A novel method, i.e. dual-invariance ESPRIT, is proposed in the paper which can estimate two directions of incident signals utilizing two shift-invariance properties of a special sensor array which consists of three shifted subarrays. The new method is very fast and can get naturally paired direction estimates. What is more, the subarray geometry is arbitrary so that accurate estimates can be obtained by using an appropriate subarray geometry.

In order to meet the applications of wireless communication, a new speech algorithm based on sub-frame joint coding is proposed in this paper. This new algorithm improves several aspects such as model structure, LSP quantization and coding. This algorithm has been implemented on TMS320VC5416 DSP processor. Informal listening testing results show that the intelligibility and naturalness performances of the 600b/s speech coding algorithm are close to those of the 2.4kb/s LPC algorithm, while the rate is only 1/4 of LPC algorithm.

In this paper, first, the frequency scatter model of radar target two-dimension electromagnetic scattering is studied, then, the idea of ISAR by parameter estimation method is presented and the MP2D method is researched, finally, the experimental result is given using simulated data and measured data of Boeing 727 airplane and demonstrated that this method could accurately extract two-dimension scattering center of radar target and is low sensitive to noise.

Modern measurement radars have to estimate the direction of the targets accurately in real time because they usually have to track the targets in real time. A novel real-time angle estimation method, which utilizes the theory of high-resolution DOA estimation and the special conditions of the measurement radar, is proposed in this paper. When compared with the conventional FFT method which is based on the principle of phase comparison, the two methods have almost equal computation cost, however, the new method has much better estimation accuracy when SNR is low.

This paper proposes a switched-beam smart antennas scheme which is based on the simultaneous searching of beams and channels, and analyzes the SIR performance of uplink in system when applying the scheme. The proposed scheme has better performance compared with the traditional respective searching scheme of singular searching of the beams and channels. It accords with the developing way of smart antennas in future mobile communication systems.

Based on the concept of contention reservation access and polling transmission, User-dependent Perfect-scheduling Multiple Access (UPMA) protocol for supporting node mobility, multihop architecture and quality-of-service (QoS) of multimedia traffic is proposed. It improves channel utilization greatly and guarantees a node to access channel rapidly. Moreover, it provides QoS guarantees for real-time traffic. Finally, the network performance of Internet data traffic is simulated and related conclusions are given.

Bluetooth is a new technology for short range wireless communication.This paper describes the characteristics of the piconet and scatternet of Bluetooth,and analyzes the basic principle of designing routing method for Bluetooth. This paper puts forward a kind of rounting technology about IP over Bluetooth. The focus is on the routing discovery and packet forwarding in piconet,in scatternet and between different scatternets. A kind of roaming method for Bluetooth unit among different scatternets is described too.

When numerical methods are used to solve the inhomogeneous media problems, generally, the large coefficient matrixes are generated. This results in the shortage of memory or very long computing time. In this paper, the domain decomposition method is applied to three-dimensional inhomogeneous media imaging. By means of dividing the domain into a few subdomains and iteratively solving the small problems on these subdomains, the large problem can be efficiently solved. The iterative curve shows the fast convergence rate. The imaging results of some complex inhomogeneous media problems are given in this paper.

The radiation response and electric characteristics of 6H-SiC MOS structure are studied with experiment. It is found that the main electronic conduction mechanism in the high field regions of the I-V characteristics is identified to be Fowler-Nordheim tunneling. The fact that ionization radiation effect becomes more notable at high oxide electric field indicates that the generation of radiation induced charge-centers near SiC/SiO₂ interface and in gate oxide under high electric field is higher than that under low electric field. SiC MOS structure has a more significant tolerance to y rays than Si MOS structure. Flatband voltage shift of SiC MOS capacitor is less than 2V under radiation of 58kGy(Si).

This paper presents a method for identity verification based on matching of hand shapes. The method first represents the shapes of hands by sets of ordered points. Next, the two sets of points are matched using point matching algorithm based on deterministic annealing and get the two matching parameters: mean matching error and matching rate. Finally, the classifier is designed for classification/verification. Considering the research of hand shape verification usually works in practical cases of limited or small samples, Support Vector Machine (SVM) is developed for verification. SVM is a new technique in the field of Statistical Learning Theory (SLT). The preliminary results show that the method can obtain higher levels of accuracy and robustness than the existing systems that based on hand geometry measurements.

Content-based image retrieval is a key technique in multimedia, network communication and computer applications, etc. In this paper, a content-based image retrieval method is proposed, which can be executed in fractal domain and does not need to do fractal coding of the iconic image. Thus the retrieval complexity and retrieval time can be reduced greatly. Experimental results have shown that the method proposed is very effective and better than the other three retrieval methods.

In recent years, the reduction of Radar Cross Section (R,CS) for antenna has been an important topic in the field of electronic warfare. The way to reduce the RCS of bi-reflective antennas, which have been widely used, in the axial direction is discussed in this paper. The conclusion shows that reasonable selections of the focal distances of the main reflector may be able to obtain low RCS, keeping the high-gain under the antenna unchanged.

Based on electromagnetic theory, the analytical and numerical results of radial energy of linear element planar antenna array excited by electromagnetic pulse are offered at first, then, by analyzing the results with the interference and the diffraction theories, following conclusions are got: (1) The electromagnetic radiation problem of plane linear element planar antenna array can be solved by grating equation concisely. (2) The maximum energy of bounded beam is proportional to the square of the number of element of antenna array. (3) Keeping the same line elements of antenna array feeding at the same time while each row successive postpone properly, the beam can be restrained within the expected area. (4) The effective scanning angle range depends on the given half angle width.

Abstract Image segmentation using deformable models encompasses a class of techniques developed in recent years and has been studied extensively. The techniques offer a novel method for effectively extracting the border of an irregular object in images. This paper introduces briefly the fundamental principle and history of image segmentation techniques. A few typical deformable models are reviewed including a discussion for their advantages and disadvantages. Finally, the limitation and the further possibilities of current segmentation techniques using deformable model are analyzed.

This paper summaries the recent research advance in the field providing congestion control mechanisms as powerful and flexible as the ones implemented by a stateful network. In a partial stateful or stateless network, three kinds of mechanisms are introduced for Internet routers: partial per-flow state, stateless core, congestion pricing and their representative implementation schemes are discussed in detail. The paper summaries this research field from the information and incentive point of view, discusses the internal relationships between these mechanisms and analyzes their performance characteristics.