Earticle

Image edge detection is an important part of image processing, and the effect of edge detection is also directly affected by image analysis, recognition and understanding. Canny operator is the most commonly used image edge detection operator. However, this operator has some limitations. The traditional Canny operator uses Gaussian filtering which may bring problems such as missing edge information and false edge. Besides, the selection of high and low thresholds of the traditional Canny operator are not accurate, and cannot be carried out by self-adaption. In order to solve these problems, this paper presents an optimized algorithm for Canny operator. In this paper, an improved anisotropic diffusion function is used to filter the image, and the improved filtering not only reduces the noise, but also maintains the edge information of the image. Additionally, this paper has improved the maximum between-class variance method (OTSU) to select the high and low thresholds of Canny operator by self-adaption. The improved algorithm is applied to edge detection of various images, and the results indicated that the improved Canny operator is effective in reducing noise and extracting edge.

The purpose of this paper is to compare, analyse and summarize the designs of existing emergency medical service ultrasound machines and evaluate the results obtained from each design to determine the most efficient and effective system. The medical ultrasound has been developed based off sonar technology and enables users to visualize the internal tissue structure of various biological organisms in real time. With the evolution and advances of analog technologies, digital processing and manufacturing, ultrasound systems can spread to different applications. In addition to this, as these technologies shrink in size and cost, they become more available to healthcare applications. Various designs for ultrasound systems are available, however the most efficient and effective design is not known. This paper seeks to compare and contrast three design choices and their produced results made in current literature and the produced results from the designs of portable ultrasound systems.

Because of the large data in the image database, the key problem of the retrieval algorithm is to retrieve the required image in the short time. Aiming at this problem, this article given a self-learning deep belief neural network method, and through building layers, input, output, and self-learning algorithm in network architecture to get global algorithm for image retrieval. The accuracy and the convergence of the proposed retrieval method are verified by experiments.

Spectrum sensing is one of the key functions of cognitive radio networks (CRN), which senses licensed band for CR users. This paper introduces a two phase detector, carries a weighted-energy detector (W-ED) and a correlated-generalized likelihood ratio test (C-GLRT). In first phase, we identify the energy and then if required, the C-GLRT makes the final decision in second phase. Performance of the proposed two phase detector is compared with the existing energy detector (ED), generalized likelihood ratio test (GLRT), and adaptive spectrum Sensing (ASS) detectors.

The deformation of the location of structural systems is an important step to predict the performance of the system under different conditions. In bridges, the plastic deformations generally occur over a long-span beam after years of service. To improve the detection accuracy and stability of bridge deformation, in this paper, we proposed a new deformation inspecting framework. The total procedure of our algorithm may be concluded as the following three aspects. Firstly, because of the beams of bridges are very long, to improve the resolution of images, we divided the acquiring of the bridge beam image into a sequential images. Secondly, the sequential images are proposed with character point extraction, image stitching, image segmentation, and the calculation of the bridge deformation. Finally, the calculated deformation was compared with the experimental results. According to the experiments on the real bridges and the simulating model, the results indicated that our algorithm may improve the detection accurate to a large extent. At the same time, the proposed algorithm is more flexible than the former proposed algorithms.

This paper presents a decolorization method using gradient and saliency as the maintained features in the conversion to preserve the local and global visual perception. First, we construct a linear parametric mapping function of RGB color channels. Then, we calculate the feature value of each pixel in the color image and the parameterized grayscale image, the feature value integrates the pixel gradient and region saliency. Finally, we search for the parameters which can get the minimum of the total differences between the feature values of color and grayscale images, and substitute into the linear parametric function to get the decolorization result. To enhance the efficiency of getting the best parameters, we properly relax the strict computation formulas of the gradient and saliency to construct a linear least square problem, and obtain the optimal parameters by solving optimization. Experimental results show that our method using the discrete searching strategy can maintain the contrasts meanwhile avoid the excessive enlargement of the contrasts during the color-to-gray conversion, this property guarantees the preserving of the visual perception. Our method using the linear least square strategy can reduce the computation time and frequently get the similar results with our discrete searching method.

Images will produce noise in the process of storage and collection. Wavelet threshold de-noising is a simple and effective de-noising method, but the choice of threshold function is a key. The hard-threshold function is discontinuous and there is the deviation between the signal processed by the soft-threshold function and the real signal, so this paper constructs a new threshold function at the origin sufficiently smooth to deal with above problems. A parameter is added to the new threshold function, which is between the soft-threshold and hard-threshold function by adjusting the parameter. The new threshold function can remove the noise effectively, and the image information is well preserved. Hence it plays an important role in follow-up edge detection. The de-noising method with improved wavelet threshold is presented, and then uses morphological edge detection on the new image in this paper. The result shows that the method can detect the complete edge effectively, and the visual effect and objective evaluation are good.

The prior knowledge of face scale and shape are introduced into active contour model for face contour segmentation. Based on the variance of each column about image and the gradient of variance, the face outline size and central coordinates are obtained. The level set functions of the collected shapes are used as training data, which are projected onto a low dimension subspace. The attribute reduction of training set by PCA method is approximated Gaussian distribution. A constructed shape prior model with facial personality traits is incorporated into variational level set model based on boundary and region to constrain the contour evolution process, and then model can be accurately evolved into the face boundary. Experimental results validate the efficiency of this method.

Image segmentation subdivided an image into its constituent regions or segments. This regions or segments of an image is known as ‘cluster’ and the method used for this is called ‘clustering method’. There are different methods or algorithms are out there to segment an image. There is a problem with those algorithms user has to supply the number of cluster in which it has to be segmented. Here we introduced a method using combination of both entropy of RGB color component and histogram to automatic determination of cluster present in a color image.

The performance of video automatic classification algorithm depends largely on the extraction of video features and selection of classification algorithm. From the perspective of video contents and video style type, the paper presents a new feature representation scheme, i.e. MPEG-7 visual description sub-combination model, a new method based on support vector machine (SVM) to solve problems with existing algorithms, by analyzing visual differences between five types of videos. Also we improve the classifier decision scheme and then propose the secondary prediction mechanism based on SVM 1-1 approach, improving the accuracy of SVM multi-classification method. The experimental results indicate that the proposed method manifests differences of different videos about feature selection, enhances the discrimination ability of videos pending for classification and increases the effectiveness of SVM multi-video classification.

As an efficient and reliable method in 3D reconstruction and active measurement, structured light vision technology is being more and more important. Structured light vision technology by one or more image coding pattern projection to measure the scenario, and with the position of the projection direction at an angle with the camera intake scene projection image, and then match the projected image and the code pattern corresponding points using trigonometric obtain a 3D scene information. In order to achieve accurate projection image and coding pattern matching, on the one hand need effective structured light coding method, on the other hand, it is necessary to use a variety of image processing method to accurately locate the position of the projection image feature points. In this paper, a color structured light coding and decoding method based on topology structured and regional structured model is proposed. Finally, the methods for extracting encoded feature points based on waveform analysis and for classifying colors based on clustering analysis are developed. The result of the experiment has shown that the proposed method has good performance for the structured light vision system.

The paper proposes a novel scheme for watermarking color images. Majority of the papers published on color image watermarking consider color image watermarking process as a normal extension to grayscale image watermarking techniques without attending to the specific characteristics of color images. In color image watermarking, the host color image is embedded using a unique mark with the sole intention of making the host image identifiable. The objective of this work is to find suitable regions in a color image where the watermark embedding could be effected robustly without significantly degrading the image and at same time ensuring the impact of embedding will remain imperceptible. The proposed algorithm uses Discrete Cosine Transform to transform the host image into frequency domain for selecting appropriate coefficients for watermark embedding. Also, instead of resorting to global embedding of watermark over host image, safer regions that are oblivious to watermark embedding are selected for watermark integration. The measures discussed ensures that the watermarking process do not introduce visible distortions. A new metric, named as GEI (Global Embedding Impact), is also defined for measuring the impact of embedded watermark information over each host image pixel.

In this paper, a simplified structure of integer lifting wavelet filter banks for lossless image compression is proposed by shifting and merging the scaling factors of the row and the column wavelet transforms. It is implemented by reducing the numbers of scaling factors and considering the scaling lifting. The numbers of scaling factors of the 2-D wavelet transform can be reduced by shifting and merging operation, and then the computing speed can be improved. Furthermore, the scaling lifting of simplified structure can be used to reduce the computing errors and get more accurate results. Experiments show that the simplified integer lifting structure results in lesser computational steps than the standard integer lifting structure and therefore improves the speed of the image compression. Besides, using the new lossless image compression system based on simplified integer lifting wavelet, the lower bit-rates are obtained.

Advancement in digital technologies has resulted in increased data transfer over internet in recent years. As a result, security of images/data is one of the biggest concern of many researchers. Therefore several cryptographic schemes have been proposed for image/data encryption. An efficient cryptographic scheme is one that have high brute force search time, low execution time complexity and should be able to provide good security. In this paper, several ciphers (traditional as well as modern) for images are compared based on various parameters such as: Time complexity, Peak Signal to Noise Ratio (PSNR), Number of Pixels Change Rate (NPCR), Unified Average Changing Intensity (UACI) and Entropy. In addition, the paper also shows the shortcomings of traditional ciphers that were used for text and how modern ciphers overcome this limitations. The analysis of simulation result shows that chaotic encryption schemes are most efficient and better than others.

According to the inherent deficiency of traditional beamformer on anti-noise and lateral resolution, this paper proposes a method based on the combination of Chirp-coded and adaptive dual focusing beamformer (CARDFB). Firstly, Chirp-coded technology is introduced to improve the transmitting average power, and then the dual focusing beamformer and adaptive beamformer are combined to obtain a higher quality beamforming signal. Finally, the signal filtered with matched filter is used for imaging. Experimental results indicate that the proposed CARDFB algorithm can achieve higher lateral resolution, better main-to-side lobe ratio and are more robust to noise compared with traditional dynamic focusing combined with adaptive focusing beamformer.

Optical networks offer higher data rates, and are fast and error free. However, non-linearity hinders them from being a perfect medium. Especially, FWM degrades the transmission characteristics of the optical systems and networks. FWM may result in crosstalk among optical channels. The paper presents a new approach to suppress the effect of four wave mixing using combination of hybrid modulator and optical rectangular filters. The performance of DWDM system and behaviour of four wave mixing is analysed. Hybrid modulator used here consists of Mach-Zehnder modulator followed by dual drive Mach-Zehnder and AM modulator. The system comprises of 8 channels each of data rate 5 Gbps. Also, comparative analysis is carried out using different fibres (Single mode fibre and ITU-T G.655). The maximum achieved Q-factor for SMF is 13.05 and 17.87 in case of ITU-T G.655.Dispersion compensating fibre is used to control dispersion. System is analysed for the length of 580 km in case of SMF and 530 km in case of ITU-T G.655.

The calibration between a camera and a two-dimensional laser scanner (2DLS) is an essential step in the object detecting system. Many algorithms with linear model have been proposed. But these tend to solve intrinsic and extrinsic calibration parameters separately and are influenced seriously by the poor initial data, which leads to unstable and inaccurate results. Hence, a new nonlinear model based on the Back Propagation neural network trained by the Levenberg-Marquardt algorithm (LM-BP) is presented for calibration in this paper. Before the calibration, the original laser data is fitted linearly to avoid the ranging error and is optimized by an angular increment to reduce the step-angular error. Then, the calibration network with 4 inputs composed of the lasers points’ coordinates and constant 1, and 2 outputs are obtained, expected values of which are the coordinates of corresponding points in the image coordinates. The sum of square of errors between the network outputs and expected values is taken to adjust the modifications of the weights and thresholds with the Levenberg-Marquardt method to optimize the calibration model. Finally, compared with related researches, experimental results show that the accuracy of calibration between camera and 2DLS is significantly improved, and the detecting system is more suitable for actual measurement situations.

Digital filters with lined phase responses that is constant group delay responses are desirable in many uses for signal treating. In this paper, a new method is proposed for scheming maximally smooth IIR filters with constant group delay in the pass band. This paper presents a concept of IIR filters with equalized group delay response. The group delay has been equalized by all pass filter. In this IIR filter design some constant group delay has been introduce and effect of group delay on phase is studied. At some extent of frequency group delay of compensated filter is linear so that phase of the IIR filter is linear at particular range of frequency. This paper looks at a compromise in which the IIR filter is followed by a allpass filter, which adaptively tries to linearize the overall phase response but leaves the magnitude response unchanged.

An image retrieval method based on the combination of visual dictionary and region of saliency was proposed in this paper, which aims to increase the accuracy of image retrieval. The image is divided into sampled blocks and the low-level features are extracted from these image blocks. Then a variety of features vectors are taken as the input vector for learning its corresponding visual dictionary respectively by non-negative sparse coding. Spatial information is added into the sparse representations of image by proposing the saliency polling method, and the similarity measure between sparse representation vectors is defined as SED (Squared Euclidean Distance), which considering the same nonzero entries and Euclidean distance of vectors at the same time. Results of experiment carried on Corel and Caltech datasets showed that this method can effectively improve the accuracy of image retrieval compared with the methods of single visual dictionary.

Bovine iris recognition technology plays a very important role in meat food traceability system about large-scale livestock individual identification. In order to improve the precision and speed of bovine iris location, iris inner and outer edge location method based on dynamic contour tracking and least square principle ellipse fitting were respectively proposed in the paper, according to the characteristics of bovine iris image. For iris inner edge location, firstly, the pupil centroid was determined, and pupil sub-image was separated from eye image, then, using dynamic contour tracking method based on the non-initialized level set model to track the boundary of pupil, and got iris inner boundary. For iris outer edge location, noise reducing template was determined by pupil sub-image, which combined with mathematical morphology operators to further suppress random noise, so, ideal outer edge detection curve was obtained. Finally, it got iris outer edge of ellipse fitting based on least squares principle. Experimental results shown that this method was less running time, high location accuracy, it has a certain practicality in large-scale livestock meat food traceability.

Magnetic Resonance (MR) images provide physicians with vital information about different diseases of the human body. Thus, such images must have adequate clarity to become highly beneficial in the medical field. However, it is known that MR images have a poor dynamic range which significantly affects their visible quality due to the deficient brightness and contrast. In order to deliver evident results, a tuned single-scale Retinex algorithm is utilized in this study to ameliorate the dynamic range which eventually results in better brightness and contrast. The obtained results are compared with various algorithms that utilize contemporary, complex and renowned concepts. Moreover, many naturally-degraded MR images are used for experimental and comparable purposes. Finally, intensive experiments revealed the favorability of the adopted algorithm, in that it produced evident results without any visible flaws and outperformed the comparable algorithms in terms of visible quality.

In this paper, the design and the analysis of the double square ring (DSR) metamaterial is presented, which is based on the equivalent circuit theory and the analysis of the distribution of the electromagnetic field. Based on the retrieval algorithm, the effect of the structural parameters on the effective permittivity and the effective permeability are proposed. The effective permittivity and the effective permeability of the designed DSR metamaterial are simultaneously negative in the range from 4.8GHz to 5.65GHz. A conventional microstrip antenna is designed as comparison, and the analysis of the microstrip antenna is not written in this paper for brevity. The superstrate is composed of an 44 array of the DSR metamaterial, and the novel antenna is consist of the conventional microstrip antenna and the superstrate. The effect of the distance between the superstrate and the antenna are simulated by HFSS software, and the novel antenna is proposed according to the analysis of the distance. The simulation results show that the gain of the novel antenna increased by 1.1dB, and the half power beam width decreased by 40.9°. Therefore, it is concluded that the performance of microstrip antenna can be improved by using DSR metamaterials.

As an important subject of image processing, image restoration is a kind of degeneration by establishing the mathematical model of reverse deduction arithmetic, and the image processing technology of the original image is obtained.Traditional image restoration algorithm by image dimension is low, the method of single factors such as limit, image restoration effect is limited.Combined with adaptive chaotic genetic algorithm, this paper proposes an improved image restoration intelligent algorithm.Through analysis and experiment comparison, the new algorithm can get better effect of image restoration.

A novel technique to generate an arbitrary chirped waveform by harnessing features of Lithium Niobate Mach-Zehnder analog intensity Modulator is suggested and verified by mathematical analysis. The most important application of chirped microwave waveform is that, it improves the range- resolution of radar. In the proposed approach, two electrical drive signal of opposite gain is used to modulate light coming out from the continuous-wave (CW) laser in LiNb Mach-Zehnder modulator (MZM). The output of the MZM is chirped optical waveform due to the dependence of the mach-zehnder modulators chirp on the form of signals applied to the drive electrode. In our simple and straight forward method of producing chirped waveform, is basically based on the concept of change in phase at the two arms of Lithium Niobate Mach-Zehnder modulator due to change in refractive index at the respective arms when electric drive voltage is applied. The chirping phenomenon is expressed in terms of an intrinsic chirp parameter of the MZM. The proposed design is simple, easy to implement and cost effective than the previously proposed model of chirped waveform generation uses the MZM and chirped fiber Bragg grating.

Aiming at the existing edge detection algorithm of edge vague, the pseudo-edge cannot be removed and algorithm results do not achieve optimal results by virtue. In order to improve the reliability and effectiveness of edge detection, the proposed optimization tool template coefficient method, to design the coding, Sobel filter and fitness function of genetic fuzzy clustering algorithm. Through interpolating, smooth handling and filtering with the updated active contour model. Based on the traditional edge detection algorithm is analyzed, combined with fuzzy membership functions and genetic operators for edge detection algorithm was improved by genetic fuzzy clustering. Through the simulation results showed that this new algorithm was feasible. Theoretical analysis and experimental results demonstrate that, the new algorithm in this paper is highly antinoise and able to get better image edges.

Multi-task Learning (MTL) algorithms aim to improve the performance of several learning methods through shared information among all tasks. One particularly successful instance of multi-task learning is its adaptation to support vector machine (SVM). Recently advances in large-margin learning have shown that their solutions may be misled by the spread of data and preferentially separate classes along large spread directions. In this paper, we propose a novel formulation for multi-task learning by extending the recently published relative margin machine algorithm to the multi-task learning paradigm. The new method is an extension of support vector machine for single task learning. The objective of our algorithm is to obtain a different predictor for each task while taking into account the fact that the tasks are related as well as the spread of the data. We test the proposed method experimentally using real data. The experiments show that the proposed method performs better than existing multi-task leaning with SVM and single-task leaning with SVM.