Earticle

An iris localization algorithm with the fast speed is proposed based on geometric features of the circle combing coarse localization with fine localization to avoid the problem of the slow speed of the classical iris localization algorithm. First, the row and column scanning method is employed to find four strings in the pupil. Four couples of tangency points are located by using the perpendicular bisector of strings and the threshold of the pupil boundary. The mean value of the coordinate parameters of the tangency points is regarded as the parameters for inner edge rough localization parameters. Then the calculus method is used for precise localization of the inner edge. The priori knowledge of the close distance between inner and outer edge centers of the circle is used to narrow the search range of the calculus method localizing outside edge, to significantly improving the speed of iris localization. The experimental results show that the proposed algorithm can improve the speed of iris localization with the high localizing accuracy.

The perfect difference codes minimize the Phase Induced Intensity Noise (PIIN) and cancel the multiple access interference. Earlier researchers analyze the performance of OCDMA systems considering three noises namely PIIN, shot and thermal noise for 3D Perfect Difference Codes (PDC) for OCDMA system. In this research paper we have considered two more noises i.e. 1/f and generation recombination noise for 3D PDC based OCDMA system. Mathematical analysis has been done and the performance of the system has been evaluated for higher data rates i.e. 1Gbps to 10Gbps for minimum BER e−9.

Through the research of the application of support vector machine theory in the pixel unmixing, the advantages of the weighted posterior probability support vector machine theory in pixel unmixing is presented. Considering the difference of each support vector machine classifier, posterior probability pixel is used as weight coefficient of sub-pixel classification for pixels unmixing. This paper presents a weighted posterior probability support vector machine mixed pixel unmixing method. This method not only has the nonlinear model decomposition characteristics of high precision, but also reduces the standard support vector machine calculating the amount of multi classifier, and it has strong adaptability.

In order to measure the dispersion coefficient of glass and other transparent material rapidly and accurately, the paper studies the dispersion coefficient measurement system based on CCD imaging automatic alignment technology. On the basis of consideration of the air refractive index impacting on the measurement accuracy of dispersion coefficient, this paper rebuilds the mathematical model by the traditional dispersion coefficient measurement formula. Using the measuring angle technology of precision grating, the paper realizes the digital measurement of deflection angle of refracted ray. Sub-pixel edge detection algorithm is adopted to realize the automatic alignment of double line to single line, which has reduced artificial error of the dispersion coefficient in the measurement process. Finally, the paper combined the multichannel small collimator designed technology and the modern control technology to achieve the high precision measurement of refractive index. Demarcating the measurement system by known dispersion coefficient of optical glass, testing results and error of standard is less than ±2.3×10-6 absolutely.

The abstract is to Recognition of handwritten Devanagari numerals has many applications especially in the field of postal automation, document processing and so on. Due to its vast applications, many researchers are actively working towards development of effective and efficient hand written character/numeral recognition. Devanagari script is widely used script in Indian sub-continent, also devanagari script forms the basis for many other scripts in Indian sub-continent. In this paper, we have proposed a hybrid method to recognize handwritten devanagari numerals. The proposed method uses, stacking approach to fuse the confidence scores from four different classifiers viz., Naïve Bayes (NB), Instance Based Learner (IBK), Random Forest (RF), Sequential Minimal Optimization (SMO). Also, the proposed method extracts both local and global features from the handwritten numerals. In this work, we have used Fourier Descriptors as global shape feature. Whereas, the pixel density statistics from different zones of the numeral to describe the numerals locally. The proposed method has been tested on large set of handwritten numeral database and experimental results reveal that the proposed method yields the accuracy of 99.685%, which is the best accuracy reported so far for the datasets considered. Hence the proposed method outperforms contemporary algorithms.

Image segmentation is a fundamental and challenging problem in image processing and often a vital step for high level analysis. Considering of the inefficient curve evolution against weak boundary and intensity heterogeneous images, an improved level set segmentation framework guided by the image gradient function is proposed. In this framework, the edges and regions of the image are roughly divided by using the image gradient sample function. Compare to the Local Binary Fitting (LBF) model, local and global intensity fitting (LGIF) model, and Edge-flow based active contour model, this algorithm may improve efficient of curve evolution in a large extent. After that, we compare this algorithm with the other active contour model to show that segmenting the noisy blurry boundary and intensity heterogeneous images can be achieved, and still go on an in-depth comparison of these models. Finally, we show the results on challenging images to illustrate the accurate segmentations.

Using Big Data in statistically valid ways is posing a great challenge. The main misconception that lies in using Big Data is the belief that volume of data can compensate for any other deficiency in data. There is a need to use some standards and transparency when using Big Data in survey research. Certain surveys that are based on the Big Data tend to generate more complications and complexities in data such as some important variables tend to correlate with some errournious data. This correlation of data with residual noise causes the endogeneity problem. It is to be solved as a fact the main aim of research work is answering question which could only be done when data is fully analyzed. Through this we can utilize all available information. This paper throws light on addressing endogeneity particularly to the astronomical data set and also provides solutions and techniques for handling endogeneity in the respective data set. Finally it couples big data i.e. whole data of sky with the time domain.

Histogram thresholding is one of the most popular image segmentation techniques. Variance-based thresholding is a famous method in which. In this paper, a new method based the framework of two-dimensional gray level histogram and class variance criterion is proposed. The methodology for image segmentation using two-dimensional histogram and variance criterion is elaborated firstly. Then the algorithm of the presented scheme is realized through recursion. Finally, the proposed method is tested on synthetic and real-world images. Experimental results show that the proposed method is better to overcome the shortcomings of the conventional variance-based methods, and the effectiveness of the proposed method is demonstrated by the experiments.

To alleviate excessive smoothing of high signal-to-noise ratio(SNR) regions which are correspond to the structural features of MR images by using classical anisotropic diffusion (AD) filter a novel AD filter called SNR controlled anisotropic diffusion is proposed. In this filter, an adaptive diffusivity function which is based on the correction factor of MR images was constructed. This adaptive diffusivity function of the proposed filter could control smoothing speed in terms of varying of SNR values in space and time scale until smoothing stop. The experimental results verify the effectiveness of novel AD filter.

Complementary filters coupled with MEMS IMU are preferred in applications where computational simplicity, low power and low cost is of prime importance. Such algorithms are equipped with fixed filter’s gain, however improvements can be realized by changing the filter’s gain as per the dynamic situation experienced by the platform. This paper is intended to evaluate the performance of conventional fixed gain complementary algorithm against adoptive gain complementary filter for attitude estimation using MEMS IMU as a test subject. As only IMU (Inertial Measurement Unit) has been exploited without using any aided sensory system, so the mandate is limited to evaluate performance of these algorithms on the basis of Euler angles roll and pitch estimation. Significant performance improvement is observed by varying filter gain in accordance with dynamic situation experienced by the system.

To highlight the saliency object clearly from the foreground, we propose a saliency detection method based on global contrast with cluster. Due to the fact that background pixels usually have similar patches, we use cluster analysis to merge the background regions. By using mean shift filter, the background pixels with similar color level are clustered and the saliency calculation can be decreased a lot. In the method, we use the contrast of color feature with all the other pixels to compute the saliency map. A weight coefficient is utilized to improve the detection accuracy in global contrast differences evaluation. The results of extensive experiments on public dataset show that our method perform well and can highlight the salient object clearly against the other five state-of-the-art methods. Besides, we demonstrate that the applications in image segmentation and fusion with our saliency map can get satisfactory results.

In poor weather condition, fog removal from an image is an unavoidable problem. Various methods have been proposed in literature for de-hazing of image based on Dark Channel Prior (DCP) such as Improved Dark Channel Prior (IDCP), Anisotropic Diffusion and IDCP method using Guided Filter. This paper provides a hybrid strategy based on IDCP using Guided Filter and Anisotropic Diffusion method. The former strategy is used for estimating and refining transmission map while the later is used to estimate and refine airlight. In addition, this paper compares the above mentioned techniques using several performance metrics such as Normalized Color difference (NCD), Color Naturalness Index (CNI) and number of saturated pixel. The results shows that the visual quality of our hybrid approach is much better as well as the performance metrics mentioned above optimizes to a great extent.

The matrix completion problem is to reconstruct an unknown matrix with low-rank or approximately low-rank constraints from its partially known samples. Most methods to solve the rank minimization problem are relaxing it to the nuclear norm regularized least squares problem. Recently, there have been some simple and fast algorithms based on hard thresholding operator. In this paper, we propose an accelerated iterative hard thresholding method for matrix completion (AIHT). Then we report numerical results for solving noiseless and noisy matrix completion problems and image reconstruction. The numerical results suggest that significant improvement can be achieved by our algorithm compared to the other reported methods, especially in terms of CPU time.

The edge detection is usually used for controlling the automation of pattern recognition in intelligent machines, traffic control and mobile devices security systems. Due to the complex of image control, the dimensions are increasing so that the single wavelet is not able to address some applications, thus, high-dimensional wavelet basis is needed. This paper introduces a construction of dual wavelet for fulfilling the gap. A dual wavelet through the compactly support set/functions which can form the Riesz basis is built up. Based on the mathematic presentation of the construction of the dual wavelet, which is a dual orthogonal with scale function and duality, the constructed dual wavelet is used for an application testing with fingerprint. The proposed approach is compared with Sobel and Canny filter. It is observed that from the results, the proposed dual wavelet outperforms the Sobel and Canny filter in terms of effectiveness and computation time.

Information security is one of most important concern in any communication. There are number of security attacks related to information security and number of techniques has been implemented to prevent these attacks. Data hiding is related field to information security. Data hiding can be achieved be using Digital Steganography. Digital Steganography is art of hiding secret messages behind the innocent looking digital media. In this paper we have done state-of-the-art survey on image steganographic techniques and steganographic techniques used with different cryptographic algorithms, to provide extra layer of Security.

The denoising of acoustic emission (AE) signal plays an important role in structural health monitoring. This paper proposes the improved spatially selective noise filtration (SSNF) which can eliminate the Gaussian white noise well. Firstly, through the comparison of different vanishing moments, “db5” is chose as the mother wavelet. And the Mallat algorithm is used in the composition and reconstruction of signal processing. Secondly, according to the signal noise ratio of wavelet reconstructed coefficients of AE signal, two coefficients are chosen to the next step. Lastly, the denoising algorithm uses the high degree of correlation between coefficients to realize the improved SSNF. Compared with the SSNF, the improved SSNF can avoid “glitches” and realize real-time denoising. And according to the simulation results, the improved SSNF can realize real-time denoising of AE signal.

Recognition of single object in Natural Images usually shows low accuracy and low comprehension ability due to the complexity of the scenes. In contrast, human vision boasts extraordinary perceptive ability, which motives our study on simulation of human visual information pathways. By mimicking the layering processing mechanism, we designed a computing model and method for image perception with the functional features of biological vision. This model can perceive single object against complex background with the accuracy above 90%, as proven in experiments based on the testing sample from the database “caltech101”.

The growing demand of securing the online trade and digital content bought the digital watermarking as major tools for protecting copyright, preventing unauthorized distribution, track back the source, confirming the ownership, authentication and many more. This paper proposes a new algorithm which shows better resistance against some common attacks. The SVD-DWT technique is applied here for deconstructing and reconstructing the image. The secret watermark is embedded into the R band using a secret key after separating the host image into three respective bands namely R, G and B. The extracting process is performed by splitting the watermarked image into all the color channels and then applying further process into R plane. The experimental result shown in tabular form indicates the better robustness of the proposed algorithm against various attacks on image.

This paper proposes a novel image compression algorithm using curvelet transform. The proposed research decomposed the original image into curvelet coefficients by using fast discrete curvelet transform, after that the different scales of quantized curvelet coefficients were selected for lossy compression by the help of cutoff threshold. The proposed method was compared with image compression method based on wavelet transform. Experimental results show that the compression performance of our method gains much improvement based on PSNR, MSE and memory size. Moreover, the algorithm works fairly well for declining block effect at higher compression ratios.

In this paper, we propose and present a novel algorithm for medical image segmentation (MIS). By analyzing the current state-of-the-art related algorithms, we introduce the multi-band active contour model based limit function to make the multilayer segmentation available. With the development of image segmentation technology, the development of medical image segmentation technology also got very big, because there is no find common, accepted effect ideal is suitable for medical image segmentation method, almost existing each kind of segmentation method has application in the field of medical image segmentation. Furtherly, with the optimized aims of being robust to the noise and avoiding the bad effluence on the result, we adopt the kernel method and new initialization curve. This model suffers from low noise robustness, and model algorithm is difficult to achieve. Integrated segmentation technology refers to two or more technology is used, combined with their own advantages, so they can on the accuracy or efficiency to achieve better performance than when using a single. A new penalty term is introduced to improve numerical stability and the step length is increased to improve efficiency. As far as the robustness and effectiveness are concerned, our method is better than the existing medical image segmentation algorithms. Experimental analysis verifies the success of our method.

This paper deals with the reduction of Total Harmonic Distortion (THD) by using Multi-pulse AC to DC Conversion technique. The three-phase multi-pulse AC to DC conversion system have a phase-shifting transformer and a three-phase source. Every such type of converter provides a 6-pulse AC to DC conversion system, that’s why in order to produce more sets of 6-pulse systems, a uniform phase-shift is required and hence with proper phase-shifting angle, 12, 18, 24, 30, 36 and higher pulse systems have been produced. The performance improvement of multi-pulse converter is achieved for total harmonics distortion (THD) in supply current, DC voltage ripples and form factor. This paper presents the discussion of multi-pulse improved power quality ac-dc converter (IPQC) configuration, comparative factors and selection of specific.

In this paper we propose one hybrid feature set filtering localization approach (HFSFLA) for iris recognition. Our HFSFLA method is different from the traditional iris localization method. Firstly we combine the advantages of both linear filtering method and non-linear filtering method, which can not only remove the noise and unwanted area but also keep the useful edge information of the iris image. Secondly, we propose feature set filtering localization to locate the iris precisely. Finally, we adopt one template matching method based on hamming distance deviation to recognize the iris information. Comparison experiments between the traditional localization method and the proposed HFSFLA are conducted on three iris databases. The experimental results show that the equal error rate and the correct recognition rate of the HFSFLA are better than those of the traditional localization method consistently in all iris data sets. And HFSFLA has high correct localization rate in the all three iris databases. It is a robust and rapid localization method.

To analysis the signal whose frequency changes as time, this article presents a new time-frequency analysis method based on short-time fractional Fourier transform (FRFT) and integration of midpoints. This method uses window functions to divide the signal to many pieces, then the FRFT is used to calculate the frequency modulation (FM) rate of each piece. A chirp signal can be built based on the FM rate and the intermediate frequency (IF) signal is obtained by mixing the chirp signal and the primary signal. The frequency of the midpoints can be estimated by the chirp-z transform (CZT), therefore the frequency of every piece can be calculated. This method avoids the impact that the quick change of frequency leads to the error of short-time Fourier transform. It also can reduce the error brought by the estimation of chirp rate. Therefore, the method can improve the accuracy of time-frequency analysis. This article also puts forward a method to search the order of fractional Fourier transform, and the computer simulation and measurement verifies the effectiveness of the proposed algorithm.

In the era of today's world verification of images plays an important role. Various image editing tools are available in the market which can make changes in image in different ways. By using these tools, we can make slight changes in the image by resizing, rotating, noise addition and stretching an image or by splicing or copy move which is difficult to detect by human eyes. The digital images plays important role in many fields such as criminal and forensic investigation, military, journalism etc. So we needed some forgery detection technique for digital image. This paper reviews techniques for pixel-based forgery detection. First is copy-move or cloning and second is fast-copy move detection. In copy-move or cloning technique a part of the image is copied and pasted into another part of the image. But this image has limitation of only shifting of copied regions. So, second technique fast-copy move detection is discussed. Its result is complex but precise. Main disadvantage of fast-copy move technique it is not being able to detect for very small region.

Photograph aesthetical evaluation has been widely investigated in these decades. The most used assessing methods are mainly classical data mining methods such as SVM, ANN(Artificial Neural Network), linear programming and so on. In this paper, we presented a method based on artificial neural network and deep learning methods which is also a hot research topic recently. We downloaded a medium and a large dataset from a well-known online photograph portal and trained on them. Results showed that the accuracy of classification was above 82.1%, which was better than all state-of-the-art methods as well as a moderate result from those methods never adopted up to now.

Vision based computer interface systems provides an easy way for human to communicate with computer or machine. At an early stage of development of these system i.e. detection and tracking of objects, some parameter or feature of human body are utilized for this purpose. Skin color has proven to be the most useful and robust parameter for detection and tracking of objects in real time. Skin color detection is basically divided into two class pixel based and region based. This paper presents an overview of pixel based skin color classifier proposed by various authors with their advantages and disadvantages. The performance assessment of these methods given by various authors is also discussed here.

Wavelet transform has many good characteristics, used extensively in image fusion. In recent years, many algorithms have been developed, but there exist some inherent defects such as image blur, burr phenomenon, zigzag boundaries and image discontinuity. In this theory, without considering disadvantages of HVS, especially fused image should preserve brightness and texture features which are the most sensitive to eye, so a new algorithm combining them is proposed. Firstly, by calculating brightness and texture metrics in different wavelet decomposition subimages. And then, by using root mean square rule to get fused low frequency and high frequency coefficients respectively. Finally, performing inverse wavelet transform by the concatenation of low frequency and high frequency to gain fused image. In order to evaluate different algorithms, the assessment metric based on HVS is adopted, which is a more comprehensive and effective measure. Experiments merging IKONOS Pan image(resolution is 1 meter) with multispectral image (resolution is 4 meter) show that the proposed algorithm is the best on brightness, contrast, texture, definition, resolution, object edge regardness of visual effect and objective metric, also verifing human visual characteristic to be considered in image fusion.

Action tracking and recognition is a challenge due to human deformation and complex scene system. Tracking-by-detection methods are used to solve appearance changes problem caused by viewpoint, occlusion, scale or deformation. Here we propose a robust object tracking and generative action recognition method. Compressive sensing is improved to track object with superpixels, and the generative structural part model is designed to be adaptive to variation of deformable object. We evaluate the method on challenging sequences. Also, we make qualitative and quantitative discussion. The results indicate the method is robust, and it is adaptive to deformable object tracking and action recognition.

Key extraction from the biometric is a new direction of encryption, which solves the safety problem of key a certain extent. Iris has rich textures, and its generated signature is longer than other biometric, and has unique advantages in term of key extraction. The wavelet packet is used to decompose effective iris region into the three layers. The third diagonal high frequency coefficient is extracted as the iris feature, and the encryption key was randomly generated from iris feature to meet the encryption demand. The United States NIST encryption standard is adopted to test the extracted key, and the experimental results show that through the 7 seed test of NIST, the generated key meets the security of encryption algorithm needs.