Earticle

This study describes a new method for wheat rows detection at the middle growth stage based on Machine Vision. The algorithm includes three steps: (i) vegetation segmentation, (ii) centers points extraction and (iii) wheat rows detection. In the first step, color images were transformed into gray-level images and Otsu’s method was used to implement binarization. Based on the fact that the corresponding center points on two adjacent horizontal scanning lines can’t have a large deviation, in the second step, we firstly extracted the initial center points on the first scanning line based on a sliding window, and then gave a small shift based on positions of the initial center points which have been extracted on the previous scanning line to extract the center points for the next scanning line. Finally, the Randomized Hough transform (RHT) method was employed to locate the wheat rows. Test results indicate that the proposed method can effectively detect the wheat rows at the middle growth stage.

Image mining is used to mine knowledge from large image databases. Image segmentation, image compression, image clustering, image classification and image retrieval are significant image mining tasks. Face detection methods are used to identify the similar faces from the large collection of facial images. It has numerous computer vision applications and it has many research challenges such as rotation, scale, pose and illumination variation. Facial expression is defined as the position of the muscles beneath the skin of the face and it is a form of nonverbal communication. Facial expressions are the expression which shows the emotions and different feelings of human beings. Different facial expressions are sad, happy, fear, normal, surprise and angry. In this research work facial expressions are classified by using the optimization algorithms. PSO with LIBSVM algorithm is proposed for facial expression classification and the performance of this algorithm is compared with the existing BAT algorithm. The results of the existing and proposed algorithms are analyzed based on the two performance factors; they are classification accuracy and execution time. From the experimental results, we observed that the proposed PSO with LIBSVM algorithm has produced good results compared to existing BAT algorithm. This work is implemented in MATLAB 7.0.

The excitations and locations of sensors in the non-uniformly spaced array affect the array performance such as sidelobe level and spatial resolution. Consequently, finding the optimal excitation coefficients and sensor positions of the array to produce a desired beam pattern with the smallest number of sensors is of great importance in practice. With the aim of reducing the number of the sensors in a linear antenna array, a novel method based on ℓp (0<p<1) norm minimization for optimizing both excitation coefficients and sensor locations of the array is proposed. Compared with the reweighted ℓ1 norm minimization (IRWL1) method, the proposed method can reduce more array sensors by optimizing the objective functions that include the measurements of peak sidelobe level (PSL) and array sparsity denoted by the ℓp norm of excitations. Numerical experiments have proved the effectiveness and advantages of the proposed method in the reduction of the number of the sensors of the linear antenna array.

In structured light 3D measurement field, when the object has smooth surface, it can form a highlight area due to the specular reflection, and the distortion of the object will make a large measurement error. In order to solve this problem, this paper use seven steps sine-phase shift combined with linear diffuser to remove highlight. Firstly, the principle of removing specular with diffuser is analyzed, then the overall design of the system is introduced, which includes 3D reconstruction and system calibration method. Finally the reconstructed experiments are carried out with ceramic plate. Experimental results show that the proposed method can significantly reduce the highlights area of reconstructed image compared with the highlights area without diffuser. The diffuser can obviously inhibited by highlights, although it can not completely remove the highlights, it plays a very important role in reconstructing specular object with more accurate and better quality.

An adaptive search order calculations are introduced to accelerate the square or hexagon movement estimation in advanced video coding. As indicated by the movement incline, a table of the versatile inquiry request is characterized. For each seeking emphasis, a superior pursuit request is determined and after that the best coordinated square can be found in the early hunt stage. Some test results exhibit the computational point of interest of the proposed enhanced calculation when contrasted with past calculations. Block motion estimation and compensation are played a major role in the video compression to reduce the temporal redundancies of the input videos. Assortment of square pursuit example is created in the writing to coordinate the pieces with decreased calculation multifaceted nature however without influencing the visual quality. In this paper, we have talked about the precious stone, square and hexagon look design with versatile request to discover the piece movement estimation. These inquiry examples are created as adaptive order square hexagon (AOSH) look calculation to locate the best coordinating piece without much considering vast number of hunt focuses. Likewise, the seeking capacity is defined as exchange off standard where, digression weighted capacity is recently created to assess the coordinating point. The AOSH seek calculation and digression weighted exchange off paradigm is successfully connected to piece estimation process with the point of enhancing the visual quality and compressive execution. The goal of video pressure is accepted for the proposed strategy utilizing three recordings to be specific, football, garden and tennis. The quantitative execution of the proposed technique and existing strategies are broke down utilizing Structural Similarity Index (SSIM) and Compression Improvement Ratio (CIR). The outcomes demonstrate that the AOSH technique got the great visual quality and compressive execution than the past strategies.

In this paper, we introduce an innovative scanning synthetic aperture radar (SAR) termed as the Hybrid Phased-MIMO SAR (HPMSAR) and investigate the system performance. The HPMSAR overcomes the limitations imposed by the conventional scanning SAR (ScanSAR, spotlight SAR, or Terrain Observation by Progressive Scans (TOPS) SAR) and presents the strong flexibility in imaging mode. Every subarray in HPMSAR is simultaneously pointed in the different directions and electronically steers the azimuth beam. This not only increases the illumination time per subswath entailing a high azimuth resolution but also covers an ultra-wide swath giving short-term time for global coverage. However, multidirectional scanning imaging will produce overlapped echoes and ambiguities on receive, then this paper presents the advanced processing scheme based on two-dimension (2-D) digital beamforming (DBF) with multichannel unambiguous reconstruction technology. The key of this method lies that multichannel processing technology in azimuth ensures receiving echoes free of aliasing by using a low PRF, and 2-D DBF could effectively suppress ambiguities and separate the overlapped echoes from different subswaths. Performance analysis compared with traditional methods shows output performance of interference suppression.

Design of fractional order analog filter, using a single Operational Transresistance Amplifier (OTRA) as an active current mode building block, is presented. The order of the proposed filter is 1.5. The theoretical results have been verified with PSPICE simulation. Obtained filter was realized using OTRA using the RC-RC decomposition technique. Frequency response for the presented filter is shown in the paper. The proposed filter offers some important features: employing single operational transresistance amplifier as an active element, lower sensitivity, insensitive to stray capacitances and parasitic resistances due to internally grounded input terminals of OTRA. Simulation results agree well with the theoretical values.

1l norm is a popular regularizer in various linear inverse problems including image processing, compressed sensing and machine learning. But the non-zero entries of the sparsity solution obtained by 1l are independent with each other, which always leads to biased result to real solution. Actually, there always exist some different correlations among those non-zero entries in an image signal domain or various analysis domains. In this paper, based on a simple observation that the non-zero entries of the sparsity vector in various image analysis domains should be also approximate when the relevant signal values are proximate, we proposed a nonlocal-approximate sparsity regularizer in analysis domains by minimizing the sum of the 2l norms of those vectors with the same nonzero pattern like signal vectors. This regularizer is applied to image denoising, edge detecting, inpainting and decomposition models successively. The numerical experiments demonstrate the effectiveness of our method in terms of PSNR, visual effect and edge preserving.

Moving targets saliency extraction is the key technology in moving target detection, tracking and recognition systems. In order to solve the problems for moving targets detection tasks, such as the background interference, target occlusion and so on, a new moving target detection method by moving saliency calculation is proposed. By the Quaternion Fast Fourier Transform, the proposed method fuses four motion features as optical flows, direction vectors and frame difference. The moving saliency map, as the result, is generated. The experimental result demonstrates that our method has better ability to handle the problems in moving object detection, such as moving information loss in complex scene. The results with better accuracy score(Cgood=0.875，Cfalse=0.028) are given by our method.

Face recognition having so many application in security, access control, gaming, aging effect, health issues, and internet communication and so on. It has been perform by using different face recognition algorithms on different field of application. When we use face recognition in real-world scenarios with unfavorable conditions such as occlusion and pose variations, illumination and expressions. Here also study the challenges during face recognition under immoral conditions such as facial expression recognition with poses variations, occlusion and lightning condition. Basically it proposes several possible future direction which is excluded from the challenges used in it. Thus, it is a beneficial developing ping point for research project on face recognitions able to be used for a practical purpose or in several ways.

Compressive tracking is considerably popular in the visual tracking community in recent years. The very strong theoretic support from compressive sensing motivates many researchers to follow and there are a wide range of compressive trackers with attractive performances. The goal of this paper is to overview some of the most recent state-of-the-art compressive trackers in the literature. First, a variety of compressive trackers are thoroughly introduced and summarized. Second, extensive analyses from different perspectives, including random measurement matrix, compressive features, feature selection strategy and so forth, aim to provide readers a good understanding of the strengths and weaknesses of different trackers. Finally, several possible future trends for compressive trackers are outlined to hopefully bring some insights to interesting researchers.

In order to meet the different data browsing requests for different users to the visual quality of remote sensing images in heterogeneous network especially under narrow bandwidth environments, an online remote sensing image progressive transmission model is constructed in which remote sensing image compression and decompression are synchronized with transmission. At the same time, a pipeline-based multi-threaded acceleration method has been proposed through solving the asynchronous problem between compression decompression and transmission to improve the efficiency of remote sensing progressive transmission. At last, an idea of retry broken downloads transmission interruption has been implemented to improve end-user interactive experience. Experimental results show that the whole processing speed has been improved nearly twice without reducing image transmission quality and the amount of data transmission was reduced evidently by using the proposed progressive transmission and real-time compression model.

Natural scene images are more susceptible to skew deformation as compared to document text which makes skew correction an indispensable step in scene text extraction. This work evaluates Murthy’s Devanagari scene word slant correction method [Signal, Image and Video Processing, 7(6), 2012] on Gurmukhi scene images. The method makes use of headline feature of Devanagari which also exist in Gurmukhi script. The headline of Gurmukhi word is found by perceiving farthest located salient points as its end-points and skew angle of headline is calculated from its slope. Gurmukhi word image is de-skewed using skew angle of identified headline. The method has been tested on 100 self-captured good quality Gurmukhi and 117 publically available Devanagari scene words with average accuracy of 62.8% and 72.2% respectively. The method has been found to be working well on few samples of defective scene words, provided actual end-points of headline are preserved. It has been observed that Murthy’s method is very simple to implement, does not require any pre-processing and give good results in wide variety of practical situations. However, this method does not work well for single character words with vowel above headline and words with identified headline parallel to horizontal axis.

Subspace partition is a common method in normal MUSIC algorithm that divides the signal covariance matrix into signal subspace and noise subspace by eigenvalue decomposition. By this method, the effect of environmental noise is curbed. However, when the signal angle interval becomes small and the signal-noise ratio reduces, some certain limitations in multiple signal estimation such as loss and confusion will be presented, which means the normal method of estimation is unable to distinguish those signals we need actually. A modified MUSIC algorithm is proposed in this paper to solve the problem. A modified part in the spatial spectrum called weighting function is introduced. Some weighted operation are given to the steering vectors when the spatial spectrum is formed, making the most of subspaces and there eigenvalues. Some simulations followed are taken to discuss the performace of the modified method. Through the analysis we can see that, under the condition of a small signal angle interval and a low signal-noise ratio, the improved algorithm could achieve satisfactory result for the DOA estimation.

Aerosol optical depth (AOD) is an important quantity parameter to study the Earth’s radiation balance, climate change and environment protection. For estimating AOD by a data mining method, the synchronized records by combing satellite observed information from MOderate Resolution Imaging Spectroradiometer (MODIS) equipment with the ground-based accurate measurements of AOD from Aerosol Robotic NETwork (AERONET) work as driving attributes and prediction targets, respectively. However, compared with the number of high-dimensional remote sensing attributes, the total number of spatial-temporal collocated MODIS-AERONET observations during a couple of years is relatively not large enough for estimation modeling. It leads to unstable feature selection subsets and drops the AOD estimation accuracy. In this paper, we propose a novel ensemble approach by aggregating multiple AOD estimators. Each estimator is modeled based on features selected from remote sensing attributes by using a subsampling strategy with instance perturbation. The ensemble approach provides aggregated retrievals of AOD with higher accuracy, while also providing an estimation of retrieval uncertainty. We conducted experiments to evaluate the empirical performance of the proposed approach on two years (2009-2011) of MODIS data over 197 global AERONET sites. The encouraging results clearly showed that aggregation of estimators modeled by multiple feature selection subsets leads to accuracy improvements and uncertainty reduction in AOD retrievals.

In this paper, we investigate the state estimation problem of nonlinear systems under the condition that the prior statistical characteristic of noise is unknown. An adaptive unscented Kalman filter (UKF) is proposed. In this algorithm, the maximum likelihood principle is applied to establish the log likelihood function with the unknown noise statistical characteristics. Then, the noise property estimation problem is transformed into the maximization of the mean of the log likelihood function, which can be achieved by using the expectation maximization algorithm. Finally, a suboptimal adaptive UKF can be obtained. Simulations show that the proposed adaptive UKF algorithm can deal with the problem of filtering accuracy declination of the traditional UKF when the prior noise statistical characteristic is unknown. The proposed algorithm can estimate the statistical parameters online.

We propose a novel image inpainting approach in which the exemplar and the sparse representation are combined together skillfully. In the process of image inpainting, often there will be such a situation: although the sum of squared differences (SSD) of exemplar patch is the smallest among all the candidate patches, there may be a noticeable visual discontinuity in the recovered image when using the exemplar patch to replace the target patch. In this case, we cleverly use the sparse representation of image over a redundant dictionary to recover the target patch, instead of using the exemplar patch to replace it, so that we can promptly prevent the occurrence and accumulation of errors, and obtain satisfied results. Experiments on a number of real and synthetic images demonstrate the effectiveness of proposed algorithm, and the recovered images can better meet the requirements of human vision.

Great effort has been put into the development of diagnosis methods for the most dangerous type of skin diseases - Melanoma. This paper aims to develop a prototype capable of segment and classify skin lesions in dermoscopy images based on ABCD rule. The proposed work is divided into four distinct stages: 1) Pre-processing, consists of filtering and contrast enhancing techniques. 2) Segmentation, thresholding and statistical properties are computed to localize the lesion. 3) Features extraction, Asymmetry is calculated by averaging the calculated results of the two methods: Entropy and Bi-fold. Border irregularity is calculated by accumulate the statistical scores of the eight segments of the segmented lesion. Color feature is calculated among the existence of six candidate colors: white, black, red, light-brown, dark-brown, and blue-gray. Diameter is measured by the conversion operation from the total number of pixels in the greatest diameter into millimeter (mm). 4) Classification, the summation of the four extracted feature scores multiplied by their weights to yield a total dermoscopy score (TDS); hence, the lesion is classified into benign, suspicious, or malignant. The prototype is implemented in MATLAB and the dataset used consists of 200 dermoscopic images from Hospital Pedro Hispano, Matosinhos. The achieved results shows an acceptable performance rates, an accuracy 90%, sensitivity 85%, and specificity 92.22%.

In modern society, visual content like images and videos is increasingly becoming a new form of media to express users’ opinions on the Internet. As a complement to textual sentiment analysis, visual sentiment analysis intends to provide more robust information for data analytics by extracting emotion and sentiment toward topics and events from images and videos. Inspired by recent works that applied deep convolutional neural networks (CNN) to this challenging problem, we proposed a framework for image sentiment analysis with a novel deep neural network called Network in Network (NIN) which intends to improve the discriminability for local patches within receptive fields. We trained our network on a dataset consisting of nearly half a million Flickr images and minimized the effect of noisy training data by fine-tuning the network in a progressive manner. Extensive experiments conducted on manually labeled Twitter images show that the proposed architecture performs better in visual sentiment analysis than conventional CNN and other traditional algorithms.

This paper conducts a research on information loss of local feature existing in the image denoising process and puts forward the method of image sparse denoising of redundant dictionary based on filtering guidance. This method utilizes bias noise (additional noise and image errors after denoised image and the corresponding additional noise deviation) for image sparse expression, and extracts the feature information of bias noise to improve the effectiveness of de-noising. In the first place, based on filtering guidance, the method carries out aftertreatment to bias noise still existing after denoise the image. And then, the method, in the basis of this bias noise, designs a new dictionary training method, and obtains redundant dictionary for image processing through self-adaption. Finally, the method extracts featured texture from bias noise image based on the dictionary mentioned above. And it takes advantage of filtering guidance in combination with featured texture extracting information and denoising image to realize image restoration. According to emulated data, the performance of proposed algorithm should be better than the selected comparing algorithm and be equipped with a better visual recovery effect.

A method of pseudo 3D video reconstruction based on 2D video sequences is proposed, which transforms the conventional 2D video into their stereo version using original image and depth-map image. The red-component map is shifted left to obtain the left-view according to the parallax information. Because the left image has holes, the depth-map preprocessing and novel hole-filling simplified algorithm are used to fill the holes effectively. The experimental results show that the proposed algorithm generates 3D visual view with the same image quality as well as faster rendering speed compared with the reference hole-filling algorithm. The method adapts to the stereo pair synthesis for 3D.

Automatic identification of various facial expressions with high recognition value is important for human computer interaction as the facial behavior of a human can be treated as an important factor for information representation as well as communication. A number of basic factors such as cluttered background, occlusion, and camera movement and illumination variations degrade the image quality resulting in poor performance for identifying different facial expressions. Moreover the identification of the automatic feature detection in facial behavior requires high degree of correlation between the training and test images. Face recognition is done by minimizing the objective function which leads to selection of optimal set of fiducial points. The method preserves the local information from different facial views for mapping neighboring input to its corresponding output, resulting in low dimensional representation for encoding the relationships of the data. The proposed method Hexagonal Descriptor Particle Swarm Optimization with Knowledge-Crowding (HDPSO-KC) overcomes from local optima and improves global search process and collaborative work. The method also covers the problem of eliminating the particles in denser regions in Pareto front distribution. The proposed methodology is validated with benchmark datasets for analyzing the performance over other methods.

In image recognition field, the fact is that the trained image classifier can not recognize the images, whose class type is not the same as the training data. To resolve this problem, a new image classifier is proposed, which is based on the class incremental extreme learning machine. The new classifier can recognize the normal images well, label them with new labels, and update itself with the new labeled data. Tested on the real-world daily activity data set, the results show that our algorithm performs well.

Electroencephalograms are the neurological signals which help in the study of various diseases. These are often contaminated with various artifacts. It is difficult to study and analysis of brain signals in the existence of these artifacts. EOG, ECG, motion and EMG are the common artifacts which cause disturbance to neurological signal. This review paper focuses on the artifact removal techniques with their features. Important parameters were taken into consideration while the study of various published papers. Strength and weakness of each paper are mentioned. This review of various papers is best of my knowledge.

The main aim of the Speech Enhancement algorithms is to improve the Quality of speech. The Quality of speech is expressed in two parameters. One is clarity, and another is intelligibility. In this paper, we proposed a method to improve the quality of speech based on computationally efficient AR modeled Iterative Kalman Filter with time and frequency mask. This approach is based on reconstruction of noisy speech signals using Auto Regressive modeled Kalman filter and further to reduce artifact noise time and frequency mask is applied to the Kalman filter output. The results of the proposed method are found to be better compared to spectral subtraction, wiener filter and Kalman filter methods.

We propose a new algorithm of privacy-preserving one-class support vector machine (SVM) with horizontally partitioned data. Every participant holds a part of data with all the data attributes. They apply the same random matrix to establish their own kernel matrix. By sharing these partial kernel matrices, we generate a global kernel matrix and establish two privacy-preserving one-class SVM models, which include the linear model and the nonlinear model. Partial kernel matrix can protect the privacy of the participants, and the global kernel matrix can ensure the classification accuracy. Experimental results on benchmark data sets indicate the effectiveness of the proposed algorithms.