MULTIPLE KERNEL RELEVANCE VECTOR MACHINE FOR GEOSPATIAL OBJECTS DETECTION IN HIGH-RESOLUTION REMOTE SENSING IMAGES

Geospatial objects detection within complex environment is a challenging problem in remote sensing area. In this paper, we derive an extension of the Relevance Vector Machine (RVM) technique to multiple kernel version. The proposed method learns an optimal kernel combination and the associated classifier simultaneously. Two feature types are extracted from images, forming basis kernels. Then these basis kernels are weighted combined and resulted the composite kernel exploits interesting points and appearance information of objects simultaneously. Weights and the detection model are finally learnt by a new algorithm. Experimental results show that the proposed method improve detection accuracy to above 88%, yields good interpretation for the selected subset of features and appears sparser than traditional single-kernel RVMs.

Crack detection using integrated signals from dynamic responses of girder bridges

An innovative approach for the identification of cracks from the dynamic responses of girder bridges was proposed.One of the key steps of the approach was to transform the dynamical responses into the equivalent static quantities by integrating the excitation and response signals over time.A sliding-window least-squares curve fitting technique was then utilized to fit a cubic curve for a short segment of the girder.The moment coefficient of the cubic curve can be used to detect the locations of multiple cracks along a girder bridge.To validate the proposed method,prismatic girder bridges with multiple cracks of various depths were analyzed.Sensitivity analysis was conducted on various effects of crack depth,moving window width,noise level,bridge discretization,and load condition.Numerical results demonstrate that the proposed method can accurately detect cracks in a simply-supported or continuous girder bridges,the five-point equally weighted algorithm is recommended for practical applications,the spacing of two discernable cracks is equal to the window length,and the identified results are insensitive to noise due to integration of the initial data.

Generalized Block Markov Superposition Transmission over Free-Space Optical Links

In free-space optical(FSO) communications, the performance of the communication systems is severely degraded by atmospheric turbulence. Channel coding and diversity techniques are commonly used to combat channel fading induced by atmospheric turbulence. In this paper, we present the generalized block Markov superposition transmission(GBMST) of repetition codes to improve time diversity. In the GBMST scheme, information sub-blocks are transmitted in the block Markov superposition manner, with possibly different transmission memories. Based on analyzing an equivalent system, a lower bound on the bit-error-rate(BER) of the proposed scheme is presented. Simulation results demonstrate that, under a wide range of turbulence conditions, the proposed scheme improves diversity gain with only a slight reduction of transmission rate. In particular, with encoding memory sequence(0, 0, 8) and transmission rate 1/3, a diversity order of eleven is achieved under moderate turbulence conditions. Numerical results also show that, the GBMST systems with appropriate settings can approach the derived lower bound, implying that full diversity is achievable.

Research on Gaussian distribution preprocess method of infrared multispectral image background clutter

This paper introduces a sliding-window mean removal high pass filter by which background clutter of infrared multispectral image is obtained. The method of selecting the optimum size of the sliding-window is based on the skewness-kurtosis test. In the end, a multivariate Gaussian distribution mathematical expression of background clutter image is given.

Automated brain tumor segmentation in magnetic resonance imaging based on sliding-window technique and symmetry analysis

Background Brain tumor segmentation from magnetic resonance imaging （MRI） is an important step toward surgical planning,treatment planning,monitoring of therapy.However,manual tumor segmentation commonly used in clinic is time-consuming and challenging,and none of the existed automated methods are highly robust,reliable and efficient in clinic application.An accurate and automated tumor segmentation method has been developed for brain tumor segmentation that will provide reproducible and objective results close to manual segmentation results.Methods Based on the symmetry of human brain,we employed sliding-window technique and correlation coefficient to locate the tumor position.At first,the image to be segmented was normalized,rotated,denoised,and bisected.Subsequently,through vertical and horizontal sliding-windows technique in turn,that is,two windows in the left and the right part of brain image moving simultaneously pixel by pixel in two parts of brain image,along with calculating of correlation coefficient of two windows,two windows with minimal correlation coefficient were obtained,and the window with bigger average gray value is the location of tumor and the pixel with biggest gray value is the locating point of tumor.At last,the segmentation threshold was decided by the average gray value of the pixels in the square with center at the locating point and 10 pixels of side length,and threshold segmentation and morphological operations were used to acquire the final tumor region.Results The method was evaluated on 3D FSPGR brain MR images of 10 patients.As a result,the average ratio of correct location was 93.4％ for 575 slices containing tumor,the average Dice similarity coefficient was 0.77 for one scan,and the average time spent on one scan was 40 seconds.Conclusions An fully automated,simple and efficient segmentation method for brain tumor is proposed and promising for future clinic use.Correlation coefficient is a new and effective feature for tumor location.

Solar flare forecasting using learning vector quantity and unsupervised clustering techniques

In this paper, a combined method of unsupervised clustering and learning vector quantity (LVQ) is presented to forecast the occurrence of solar flare. Three magnetic parameters including the maximum horizontal gradient, the length of the neutral line, and the number of singular points are extracted from SOHO/MDI longitudinal magnetograms as measures. Based on these pa- rameters, the sliding-window method is used to form the sequential data by adding three days evolutionary information. Con- sidering the imbalanced problem in dataset, the K-means clustering, as an unsupervised clustering algorithm, is used to convert imbalanced data to balanced ones. Finally, the learning vector quantity is employed to predict the flares level within 48 hours. Experimental results indicate that the performance of the proposed flare forecasting model with sequential data is improved.

Semi-LDPC Convolutional Codes:Construction and Low-Latency Windowed List Decoding

This paper presents a new coding scheme called semi-low-density parity-check convolutional code(semi-LDPC-CC),whose parity-check matrix consists of both sparse and dense sub-matrices,a feature distinguished from the conventional LDPC-CCs.We propose sliding-window list(SWL)decoding algorithms with a fixed window size of two,resulting in a low decoding latency but a competitive error-correcting performance.The performance can be predicted by upper bounds derived from the first event error probability and by genie-aided(GA)lower bounds estimated from the underlying LDPC block codes(LDPC-BCs),while the complexity can be reduced by truncating the list with a threshold on the difference between the soft metrics in the serial decoding implementation.Numerical results are presented to validate our analysis and demonstrate the performance advantage of the semi-LDPC-CCs over the conventional LDPC-CCs.