Earticle

Based on the gray features and shape features of objects, some satisfied objects are detected by using sliding window algorithm from satellite image. To further recognize their identification and classification, more texture features of them are needed to obtain to compare between them. GLCM (Gray-Level Co-occurrence Matrix) statistics are used to representative each partition of them. These PGLCM (Partition-GLCM) statistics can combine into a feature vector and those detected objects can be accurately recognized and classified by using GLVQ (Generalized Learning Vector Quantization) Neural Network algorithm. Experiments show when we choose those adapted parameters, such as the length and width of the window, and the threshold of difference of adjacent pixels, the extraction rate of building objects is up to 76.1%. Using the classification algorithm based on the feature vector generating by the statistics of PGLCM, the recognition rate of building is more than 88.9%.

Digital Image Processing is basically the implementation of a set of computer algorithms for processing digital images. Digital image processing has probative advantages over Analog image processing. In this content, noise like, Gaussian, Salt and Pepper, Speckle and Poisson, is added to an image and then the original edges are restored using various filters and tools. These tools have remarkable alteration on the image and hence they are widely employed. Some of them are wavelet transform, median filter, Weiner filter and many more.

The subject of this paper is the segmentation of multiple objects from images based on 6the parameter active contour model (PACM). After analyzing application of the parameter active model to segment multiple objects, the evolution strategies and disadvantages of existing methods are presented. This paper proposes that the key points are two parts in detecting multiple objects with the PACM in the shrinking strategy. One key point includes the split time where contours appear as self-crosses, and the split algorithm of contours. The other key point is to maintain the uniform distribution of sampling points on contours in order to match the shapes of objects in segmenting. A new algorithm for detecting self-crosses is presented, and the results show that the new algorithm is faster than the other algorithm. The problem where vertexes on contours are sampled to match the shapes of objects in segmenting is studied, and its solution is presented.

With the development of modern high precision processing technology, precise microminiature electromechanical products have been widely used. Most of the precise microminiature electromechanical products are assembled from microminiature parts.Due to dimensions of the microminiature parts involved are in micro-and nanoscale, which cannot be reached by ordinary cameras, it cannot effectively capture the image of those parts to be assembled, thus affects the assembly. A microscopic vision measurement system of stereo microscope that combined with high resolution CCD camera was developed based on micro machine vision, which uses the MATLAB software to deal with the collected images of microminiature parts such as image smoothing, image sharpening, image segmentation and edge extraction, etc. to study the structures, identification, location and assembly process of microminiature parts. The results show that the visual system we used has accomplished the automatic measurement and precise assembly of miniature parts excellently.

The Z source inverter is a novel power conversion topology that can buck and boost the given input voltage. Voltage source inverter (VSI) and current source inverter (CSI) have some common problems. The proposed work eliminates the limitations of both voltage source inverter and current source inverter by replacing multilevel inverter. This paper presents the new inverter topology based on combination of Z Source Inverter (ZSI) and Multi Level Inverter (MLI). The Z source inverter employs a unique impedance network to couple the main circuit to the power source. The basic structure includes one DC voltage source, Z network and multilevel inverter. Thus providing unique features that cannot be provided in both traditional voltage source and current source inverter. The cascaded MLI composed of eight switches to generating five voltage levels. The presence of multilevel inverters provide high output voltages with low harmonics without the use of transformers. The Z source concept can be applied to ac-dc, dc-ac, ac-ac and dc-dc power conversions. This work observes the rms (Root Mean Square) output voltage and presence of THD (Total Harmonic Distortion) in the output waveform by simulating the Z source based multilevel inverter using MATLAB/simulink.

In order to evaluate the damage of the road accurately, a Ferret subsection measurement method based on Invariantmoment was proposed. First, it set a Ferret minimum external rectangular on the basis of the barycenter and spindle. The rectangular width was the result of crack width. It set the edge conditions for reducing the errors of measurement caused by the slender burrs. It allowed rectangular edge and targets to intersect. In order to section reasonably, a method based on distance map was proposed. In the extreme value points, the subsections image was finished. It ensured that the distance between two adjacent lines was longer than the crack width. Experimental data indicated that the percent of valuable information was higher, the measurement was accurate. It can meet the requirement of practical engineering.

At present, people get less and less exercise. This virtually led to a decline in the quality of people's physical. The best way to solve this problem is to develop a set of easy to install, inexpensive sports teaching system, so that people can work in a better way to get sports guidance. In view of this, this paper studied physical education teaching system, and the system has the function of 3D human motion capture, 3D human motion reconstruction and three-dimensional (3D) human motion analysis. First of all, this paper uses the Kinect platform to achieve three-dimensional human motion capture, and it focuses on the principle of 3D motion capture and technological development of 3D motion capture technology based on Kinect; Then, the motion data storage and 3D motion reconstruction of OGRE graphics rendering engine are described in this paper; Finally, this paper makes a research on the method of 3D human motion comparison, and designs the method of dynamic time warping (DTW) based on the problems existing in the actual situation, at the same time, this paper carries on the verification experiment. The experimental results show that: the three-dimensional human motion capture based on Kinect platform can meet our requirements for human motion capture accuracy. By comparison, the OGRE (Object-Oriented Graphics Rendering Engine) graphics rendering engine is more suitable to complete the rendering of the character model in the motion reconstruction. Most of the action comparison analysis method can only carry on the qualitative analysis to the movement study quality, but cannot carry on the quantitative analysis. And DTW algorithm can be a good solution to the above problems, and the physical education teaching system has high application value.

For improving automation calibration precision of multiple view texture collection in texture reconstruction, in this paper a new kind of calibration template using statistical distribution of surface texture is proposed. Firstly, there is a normalization processing be made according to statistical distribution of texture feature, such as species of color, quantity of geometric pattern and bump texture, etc. for the division of viewpoint. Then, the camera views of calibration template are assigned by using the normalization data with the calibration algorithm. Finally, the algorithm analysis is made by viewpoint division of four cases and the accuracy of our method can improve more than 10 percent. The experimental results show that, the proposed new calibration template method can better improve precision of 2D-3D registration and texture accuracy of reconstruction model than the uniform calibration template, while no obviously increasing the amount of calculation.

Reliability is very important in wireless network since large number of wireless standards are widely used in our daily life. The network flow ratio and workload will increase significantly as well. It is observed that network manager is not able to ensure the reliability of the network even if the network connection is smooth. This paper proposes a network reliability evaluation model using factorization approach for the small and medium-sized wireless communication network. The factorization approach is a decomposition of an object into a product of other objects or factors, which when multiplied together give the original for example a number a polynomial or a matrix. With the model, the solution algorithms are proposed to work out the corresponding defined objectives. Experiments show that, the proposed model outperforms the ergodic method which uses large number of loops to obtain the network reliability. From the experiment, when Pc = 0.9 and Pm = 0.1 as well as Pc = 0.9 and Pm = 0.01, it could be find that there are six transmission lines which are with the maximum reliability.

Iris recognition is one of biometric techniques that provide fast and accurate for human identification. This paper studies four main types of iris (Jewel, flower, stream, and shaker). And also it studies a fast method using Speed Up Robust Features algorithm (SURF) for finding a match between original eye image and input eye image to the system with taking into account the speed of algorithm implementation and its compatibility with an iris scanner.

In view of the greater changes of posture, illumination, expression and scene in reality environment have a strong impact on wild face recognition algorithm to identify performance problem, and puts forward a kind of linear discriminant analysis side information (SILD) algorithm on hyperplane fusion of learning prototype. First of all, using support vector machine (SVM) to weak tag of data-concentrated sample is expressed as the middle-level characteristics of prototype hyperplane, using a learning combination coefficient to select sparse support vector set from untagged conventional data set; then, under the constraints of the combination sparse coefficient of SVM model, by using Fisher linear discriminant criterion to maximize discriminant ability of untagged data set, and using the iterative optimization algorithm to solve the objective function; in the end, using SILD for feature extraction, cosine similarity measure to complete the final face recognition. In two general face data sets of wild face recognition (LFW) and YouTube, it makes comparison of PHL+SILD method and low-level features + SILD method on some characteristics, such as strength, LBP, Gabor feature and Block Gabor feature, average accuracy, area under the curve (AUC) and entire error rate (EER). The validity and reliability of the proposed algorithm is verified by the experiments.

In order to prevent the moving vehicle shadows from being wrongly detected as the target, a shadow detection algorithm fusing chromaticity, brightness and edge gradient information is proposed in this paper. Specifically, shadow feature image is established according to the variation proportions of the chromaticity and the brightness of the foreground and the background of the moving target, and then the area with the maximum chromaticity is regarded as the vehicle search start point to gradually absorb the surrounding areas with the richest edge gradient so as to form the vehicle body area, and then remaining area of the foreground not containing the vehicle body is regarded as the shadow candidate area, and then the region growing method is adopted to obtain various shadow sub-areas for integration so as to form the vehicle shadow area. The experiment result shows that this method has the advantages of little manual intervention, high shadow detection rate, etc.

Traditional image processing and pattern recognition research aimed at identifying the target image. As time goes by, on the basis of the recognition of the image, more and more research points to identify multiple targets in the image, and the corresponding block of the corresponding target is calibrated. Compared with the traditional image recognition, the problem of image annotation is a combination of multi classification and multi regression, which is more difficult and challenging. On the basis of deep convolutional neural network, this paper studies the image annotation algorithm based on region selection algorithm and support vector machine, and the algorithm is tested on the PASCAL VOC 2010 image data set. Experimental results show that compared with the current algorithm, this algorithm can be used to mark the image of multiple targets, the effect is obvious, and there is a great practical significance.

The spectrum of the random space vector pulse width modulation (SVPWM) strategy is extremely complicated due to the random variable. A new algorithm based on the Monte Carlo method is proposed to optimize and customize the frequency spectrum of the random SVPWM strategy. A universal theoretical spectrum computation method is given for the SVPWM strategy firstly. In addition, the key procedure of the proposed algorithm is presented. Finally, several computation examples are provided to verify the effectiveness and feasibility. The analysis and computation examples show that the proposed algorithm has several advantages, and the results verify its convenience and feasibility.

The paper proposes an information hiding technique for steganography and digital watermarking using Bi-dimensional Empirical Mode Decomposition (BEMD) and Hilbert Transform (HT). We use RGB color image as a cover image. The cover image is divided into a number of blocks. Then each block of the cover image is putrefied into a number of Intrinsic Mode Functions (IMFs) using BEMD. We embed secret information with private key into a significant IMF. The embedded secret information, called watermark, is a matrix of 1 and -1. However, this watermark is generated by mapping pseudo-random numbers. Therefore, any unauthorized person who does not possess the private key cannot extract the watermark. The significant IMF is a highest energetic IMF which is selected according to the energy distribution of the putrefied IMFs. Thus the selected IMF is less sensitive to common image and signal processing manipulation. In watermark extraction process, we transform the watermarked object into frequency domain using HT bypassing the use of BEMD due to its empirical characteristics. Thereafter, we extract the watermark bit using block averaging technique. The experimental results of this algorithm demonstrate that the proposed method has better imperceptibility and it is more robust against several image processing and geometric manipulations.

Position information is the foundation of massive applications in Wireless Sensor Network(WSN). Three improved positioning algorithms based on DV-Hop are proposed in order to enhance the positioning accuracy of wireless sensor nodes. First improved algorithm is distance compensation algorithm (DCA) that creates a triangle model to compensate the estimated distance. The second improved algorithm creates a new chain table for all anchor nodes to record and compute the average hop distance. The third improved algorithm is weighting different anchor nodes with anchor nodes’ nearest unknown nodes. The second and third improved algorithms are based on the DCA. The simulation results show that the three improved algorithms are better than the original DV-Hop in localization accuracy. Compared to the original DV-Hop algorithm, the simulation results shows that the three improved algorithms proposed in the paper increase the positioning accuracy of the unknown nodes.

In this paper, we propose an adaptive sampling technique that achieves considerable energy savings while maintaining the required intelligibility level for mobile sign language video communication. The adaptive sampling scheme adjusts the sampling rate of the camera sensors dynamically according to the properties of sign language video communication and available battery power of mobile devices. Experimental results show that this adaptive scheme performs better than continuous sensing scheme in balancing the energy- intelligibility tradeoffs.

Orthopedicians often identify imaging modality visually out of their experience. To be effective, the process needs to be automated. This paper presents a behavior of wavelets in classification of orthopedic imaging modalities using Artificial Neural Network (ANN). In this work, we have considered orthopedic imaging modalities, namely, X-ray, CT and MRI and Bone scan images. Four wavelets, namely Haar, Daubechies, Symlets and Coiflets are used for sub band decomposition and their approximation co-efficients are recorded. Features, namely, mean standard deviation, median, variance and entropy is drawn from the decomposed images. Results are drawn from the performance of these wavelets at five levels of decomposition. Feature reduction is based on the classification accuracies which are analysed using wavelets. The experimental results show that the proposed method achieves satisfactory results with an average accuracy of 98% for four wavelets and for all the modalities considered. The study can be extended to include other modalities in medical field. The work is useful for orthopaedics practitioners.

As more and more insight into the human brain, EEG is not only for application, processing and analysis of EEG in science and engineering field, it will also be very important in the physical, psychological and pathological studies in humans. The EEG brain - machine interface technologies to achieve the human brain and the computer or other human interface devices to communicate and control, it can provide a special kind of information exchange and live entertainment, is also disabled and a new way of control. In this paper, the collection of information preprocessing electroencephalogram (EEG) signals is proposed wavelet packet transform feature extraction method, using SVM classifier, classification based on the operation mode to achieve recognition of the EEG signal. For different individuals turn, the next turn, show fist, fist four kinds of hand motion recognition experiments show that the average recognition rate of over 80%, significantly better than the other methods to identify results.

With the development of artificial intelligence and pattern recognition technology, more and more research related to human face is constantly developing in all walks of life. At the present stage, the traditional face recognition algorithm based on LBP and SVM is not good, and the process of feature extraction and feature classification are deeply studied in this paper. For feature extraction, the authors put forward an improved CS-LBP texture feature; for feature classification, the author uses the histogram intersection (HIK) kernel function to classify the features which has high efficiency and good effect. Subsequently, experiments are carried out on the Yale data set and the ORL data set. Experimental results show that the proposed algorithm has a significant improvement on the face recognition effect of face direction change, and the illumination change is slightly improved. In the natural environment, most face recognition has the influence of human face direction and noise, and the effect of noise is a hot direction of face recognition research in the future.

In nature, the observed Chaos phenomenas were often mixed with noise, the existence of noise made the prediction of chaotic time series generate large errors. Chaotic time series had the characteristic of broadband, which liked noise. So there were some limitations with the traditional method of de-noising. But the wavelet threshold de-noising method had the characteristic of the multi-resolution analysis, and its computational quantity was smaller and the noise filtering effect was better. On the other hand, for different types of signals, with different wavelet base functions and threshold rules, it might have a different effect on the de-noising effect. In order to search for the optimal selection of those parameters, firstly this paper constructed a simulated Lorenz noisy signal, and used this signal to do the de-noising experiment, used the SNR and RMSE as the evaluating indicator, and finally obtained the matching combination of those parameters. At the end of this paper, the de-noising simulation was carried out using China's Shijiao station runoff time series data from 1960 to 1970 in China, and the final results showed the effectiveness of the proposed method in this paper.

With the purpose of solving the issue that there are errors in the traditional Loupas Algorithm used in ARFI and the effects of imaging are not as good as expected, this thesis will talk about an improved method to solve this proplem.Based on Loupas Algorithm, it is presented that we can get more accurate computing result of the displacement by changing the size of the sampling time windows and calculating the average speed according to the 2D autocorrelator.The method presented in this paper is aimed at doing computing refering to the tiny displacement caused by the push from ARFI and comparing between the displacement-time graph of the focus of the windows with different value.Finally,from these,people can know the most suitable value of the sampling time windows.Compared with the result of the experientunder traditional Loupas Algorithm,that of this experient has improved a lot.As a result,not only Elastographic Contrast-to-Noise Ratio has been improved,but also the image quality.

An image stitching algorithm based on the robustness of principal component analysis (RPCA) is proposed in an effort to suppress the influence of noise in the image stitching quality. This algorithm represents high dimensional feature data by utilizing a lower dimensional linear subspace, and converts the image stitching problem into a principal component matrix matching problem. Through the use of a low rank matrix, the extraction of salient image characteristics is recovered and the noise interference is reduced during the enhancement process. Together, with the advantages of the RPCA algorithm, the algorithm improves the PSNR of the image while maintaining its strong matching ability. Experimental results show that the proposed scheme is able to significantly inhibit the noise and improve the stitching quality in comparison to the other existing stitching methods.

With the development of embedded technology, the embedded image processing algorithms are becoming more and more complex. Edge detection is the basic algorithm of image processing, which is used in the extraction of information from the image. Laplacian operator is useful for the edge detection. Especially, the operator can be determined by the zero crossing point of the two differential positive and negative peaks. In order to meet the fast hardware implementation of embedded image processing, this paper presents the hardware acceleration of Laplacian operator by Zynq-7000. Using HLS method, the complex image algorithm is automatically translated into hardware language, and the function of the algorithm can be quickly completed. The experimental results show that, the embedded hardware accelerated image based on Zynq-7000 can not only realize software programming and hardware programmable combination, but also improve the embedded system flexibility, scalability, and accelerate embedded image processing product design time.

Three dimensional computer animation and motion simulation is an important means of technical analysis and diagnosis of sports training techniques. Springboard diving simulation system is researched in this paper by using three-dimensional computer assisted animation, and the human body model suitable of the system is established. Then we used free deformation method based on NURBS deformed to achieve the body's joint movement. Three-dimensional human body model and its motion model are established in the computer by introducing computational geometry. At last, the simulation system is realized by using OpenGL graphical programming interface, and the position difference simulation of the platform is carried out in the system. The visual effect and real-time motion of the human body model is performance good in this paper, and it has a high application value to the guiding practice.

Hyperspectral image classification has received an increasing amount of interest in recent years. However, when representing pixels as vectors, the dimensionality of feature space is high, which causes “curse of dimensionality” problem. In this paper, in order to alleviate the impact of above problem, a manifold sparse coding method is proposed. Firstly, matrix decomposition technique is used to find a concept set and calculates relative data projection in the concept set. Secondly, manifold learning regularization is imported into objective function to capture the intrinsic geometric structure in the data. Finally, LASSO regularization is used to obtain sparse representation of data projection. Experimental results on real hyperspectral image show that the proposed method has better performance than the other state-of-the-art methods.