The performance of the target tracking in video sequences was unstable only using the color information.In order to track the interested targets robustly in the conditions of cluttered backgrounds and illumination changes,an efficient Particle Filter(PF) tracking algorithm combining the color and structural information was proposed.A structural model was developed by using target gray level image,and structural similarity index was used to measure target similarity between two frames in correlation tracking and applied in color histogram-based particle filter framework.The experiments show that our algorithm can track the targets more robustly than traditional color histogram-based PF tracking algorithm when using the same particles,and can be applied in suspicious target tracking in specific situation.

To find the quantitative relation of temperature and the absorption cross-section of SO2,the SO2 with the concentration of 10 428.57 mg/m3 is detected at temperatures of 20 ℃,40 ℃,60 ℃,80 ℃ and 100 ℃ in the wavelength range of 295 nm to 315 nm,which is often used during the measurement of SO2 absorption spectroscopy.Through qualitative and quantitative analysis on the experimental result,the quantitative formulas between temperature and SO2 absorption cross-section at different wavelength are deduced.Moreover,a new approach is put forward to decrease or compensate the error induced by the temperature.

In order to study light scattering and reduce scattering losses from multilayer dielectric high-reflection films,vector scattering theory and relationship of total scattering losses based on vector scattering theory and Bidirectional Reflectance Distribution Function（BRDF）were concisely presented.Total scattering losses changing with incidence angle and polarization from dielectric high-reflection multilayer films in fully correlated and uncorrelated model were researched,respectively.Furthermore,the influence of incident wavelength on light total scattering losses was analyzed.The theoretical results show that total scattering losses of p-polarized scattered light due to p-polarized incident light(Spp) strongly depend on the correlation between interfaces of multilayer films,especially near the Brewster angle.Moreover,total scattering losses changing with wavelength are consistent well with reflection spectrum of high-reflection multilayer films at fully correlated model.The result is reverse to uncorrelated model.

Cu/Zr nano-multilayer films are prepared by Direct Current(DC) magnetron sputtering in the silex substrate,and the box-type furnace is used to anneal.Micro-sclerometer,X-ray Diffractometer(XRD) and Scanning Probe Microscope(SPM) are used to analyze the mechanical performance and microstructure of Cu/Zr nano-multilayer films.The relation of process parameters and performance is investigated.Results show that:When Cu modulation thickness varies between 42 nm and 140 nm,the microhardness of Cu/Zr nano-multilayer films gradually reduces from 4.9 GPa to 4.0 GPa;When Zr modulation thickness is from 3.2 nm to 4.8 nm,the hardness decreases from 6.51 GPa to 5.64 GPa,hardness value decreases 13.4%,then Zr modulation thickness is from 4.8nm to 8nm,and microhardness tends a steady value(about 5.7 GPa);When annealed suitably,the superficial appearance and microhardness of Cu/Zr nano-multilayer films get better.The results accord with the model of hardness enhancement which is proposed by X Chu and S A Barnett.

Least Squares Support Vector Machines(LS-SVM) are used to learn the relationships between the image information and the 3D information,which doesn't need to confirm internal and external parameters of the camera.In the case of binocular vision,the location relationship between the two cameras is implicated in the map relations.According to the LS-SVM and camera calibration characteristic,a new method of camera calibration for stereo vision based on the LS-SVM is proposed.The image pixel coordinate which gathers from the camera is made as input,the world coordinate is be used as output,and the given mapping relation of network realization is completed by the LS-SVM.The comparison with BP neural network shows that the calibration accuracy can be improved by using the LS-SVM.The real-time performance is good and response velocity is quick.

The error propagation model in structured light system was presented.Based on Taylor decomposition and the principal of matrix turbulence,the measuring error of structured light and error from light-plane calibration were analyzed,in which cross-ratio invariant was used in calibration of structured light system.The demands of precision in the specific example were presented,which were ones of calibration parameters and samples under the conditions of acquisition of ±0.5 mm precision in the measuring range of 1 000 mm.According to the error propagation,experiment was made by using the precision of calibration samples.The measuring results are that the value of Root Mean Square(RMS) is 0.333 4 mm,which meets the requirements of theoretical precision.The experimental results validate that those error analyses would provide the principle basis for wider application of structured light system.

In order to diagnose the polarization of the x-ray line emission of plasmas excited by Z-pinch,a X-ray line polarization spectrometer is developed for measuring the spectrums.With Mica(002) crystal as dispersive element,Bragg angle is in the range of 30°～60°.X-ray lines radiated by plasma are measured in the vertical and horizontal directions,respectively.The X-ray spectra of lines in Ar XVⅡare photographed by using the polarization spectrometer with X-ray film on Yang accelerator.We obtain the resonance line w,the magnetic quadrupole M2 line,the intercombination line y and the forbidden line z in the vertical and horizontal directions.The experimental spectra recorded by the spectrometer are apparently different in the vertical and horizontal directions.The degree of polarization of lines w,x,y and z is calculated to show anisotropic property and the presence of beams in Z-pinch plasmas.

The measurement principle for specular and mirror-like surface was discussed and the main algorithms and key points for shape reconstruction via gradient data were analyzed in general.The simulation and the experiment results demonstrated the precision of cross path integration techniques,Fourier transform integration and area wave-front reconstruction method.And the results prove that the area wave-front reconstruction method is not only very immune to high frequency noise,but also has excellent performance when dealing with complicated multi-connectivity surface and non-grid gradient data,which is more suitable for wave-front detection based on fringe reflection.

Camouflage coating plays an important role in the study of infrared stealth technology.Based on the infrared stealth mechanism,a new study orientation is put forward for the infrared camouflage coating technology with the judgment criterion of infrared stealth,in which the radiation contrast between target and background should be less than 0.02.Lock-on range of detector system was also used to check this criterion.The results show that it is not suitable to only reduce the emissivity of target when considering infrared stealth of target.On the basis of the radiation energy difference between the target and background,the camouflage coating with proper emissivity can be chosen for target by taking specific combat environment into consideration.It is also found that as the contrast diminishes to 0.02,the lock-on range reduces by 71.4%.

In view of limitation in correlation detection algorithm of weak and small infrared target,the combined correlation detection algorithm is proposed,which synthesizes the merits of correlation detection algorithm based on the Detect-before-track(DBT) and correlation detection algorithm based on the Track-before-detect(TBD) as two algorithm cascade method.First,the correlation detection algorithm based on TBD is used to form a transition frame,and then the correlation detection algorithm based on DBT is used to further detect.In certain detection time,the algorithm can make full use of target space and time information to further enhance target detection ability in the small and weak target infrared sequence with low signal-to-noise ratio.Algorithm performance analysis and experiment results prove that the validity of the algorithm in the condition of low signal-to-noise ratio.

Free running semiconductor Laser Diode Array(LDA) is subject to wide line width and peak wavelength drift.In order to overcome the drawbacks,an External Cavity Laser Diode(ECLD) is constructed by adopting a Volume Bragg Grating(VBG).The experiments show that,after using the VBG reflector,the output spectral width is reduced to 0.96 nm,which is narrower than its original free running value of 2.3 nm.In addition,within the variation range of the coolant temperature(14～31 ℃),the peak wavelength of the VBG external cavity LDA is stabilized at the Bragg wavelength(808 nm) of the VBG,and the spectral width is kept below 1.14 nm.Experiments also show that the peak wavelength and the line width of the VBG external cavity LDA have no observable changes when the drive current of the LDA is varied from 7 A to 13 A.

A serial of Er2O3-doped bismuth-boron-germanium glasses with composition of Bi2O3-B2O3-GeO2-Na2O were prepared by the high-temperature melt-quenching method.And the absorption spectra and fluorescence emission spectra were measured and investigated as a function of erbium ion doping concentration.It is found that there are intense radiation trapping and concentration quenching effect in erbium-doped Bi2O3-B2O3-GeO2-Na2O glasses.With the increase of erbium ion doping concentration,the 1.53 μm fluorescence emission spectrum broadens and the sub-emission peak at 1 560 nm increases,which is analyzed qualitatively with a four-level equivalent model.At the same time,the fluorescence lifetime of 4I13/2 level and the fluorescence intensity at 1.53 μm increase at first,and then decrease with the increase of erbium ion concentration.On the basis of the energy migration model of Dexter theory,the critical distance R0 for concentration quenching happening and the interaction microparameter CEr-Er were determined.The value of CEr-Er in the Er2O3-doped Bi2O3-B2O3-GeO2-Na2O glasses was compared with that in other glass hosts.

An approach for estimating three-dimensional motion field of cardiac vessel axes recovered from X-ray angiographic sequences based on elastic registration is proposed.Displacement vector of each skeleton point is obtained by matching vascular axes at consecutive instants in point-by-point manner.A proper cost function is minimized to acquire the global optimal match by using the algorithm of dynamic programming.The accuracy of the method was quantitatively assessed by using computer-simulated data with known transformation parameters.Experiments with clinical data were also carried out.Results show the validity of the method.

As the time-frequency characteristic of single wavelet base is difficult to match with complex image feature,the image denoising effect could hardly be improved further in the wavelet threshold denoising method.A new image denoising method based on linear combined wavelet base is proposed.Firstly,a new wavelet base is constructed by linear combination with multiple differences orthogonal wavelet base,and then the combined wavelet base is used to decompose the image.Finally the denoising image is obtained by threshold processing.The combined wavelet base can match well the image feature with adjusting combined coefficients to further enhance the image denoising effect.The experimental results indicate that the denoising effect of the method is superior to the single wavelet base method with 3.5 dB improvement of Peak Signal-to-noise Ratio(PSNR) at most.

Morphology is used for edge extraction of binary image.In particular,Discrete-time Cellular Neural Network(DT-CNN) is introduced into binary morphology owing to the natural corresponding relationship between them,and then the morphology operation is transformed into multi-layer DT-CNN under some specific template.Multi-layer DT-CNN is transformed into Monolayer DT-CNN by optimizing design to reduce the complexity of the operation.Therefore,we can extract the edge of gray image by synthesizing the edge detection result of each bit in the pixel.Compared with the traditional Sobel and Log operator,the method has a better performance and converges quickly.

For the sake of accelerating the coding process of JPEG2000 and avoiding coding useless codes,a pre-coding Rate-distortion Optimized(RDO) rate allocation algorithm was proposed.At first,a formula for minimizing the total distortion in a limited rate was proved by analyzing the rate-distortion model of wavelet coefficient,and then a realizable algorithm was provided following the distortion minimum formula.The experiment results indicate that the proposed algorithm realizes accurate pre-coding rate control and minimizes distortion.Furthermore,it makes the encoder just encode the bytes that are needed,accelerates JPEG2000 encoding process and reduces the encoder’s memory consumption.

we develop a novel coarse-to-fine method for the registration of the reference and target images,involving the 2D rigid transformation.The proposed method consists of four main steps:utilizing multi-scale Harris corner detector to extract the feature points in the reference and target images;corresponding these points by a novel feature points matching algorithm based on voting strategy,which efficiently eliminates the false and unstable matches;estimating transformation parameters with the matched feature points;refining the transformation parameters with the window-based method.Experimental results demonstrate that the proposed method is able to achieve high accuracy and robustness against noise with low computational burden.

To realize the stereo matching of natural scene contour edge,a fast matching algorithm based on region adjacency graph was proposed.Firstly,on the basis of image segmentation by using the watershed transformation method,the contour edges were detected according to the boundaries of the segmented regions.Then the matching process was divided into two steps based on the global-to-local hierarchical matching idea.The first step was region edge matching by grouping the contour edges according to their corresponding regions.After establishing the region constraint based on the region characteristic,the region edge matching was accomplished in the guidance of corners in a similar way of region growing.The region constraint reduced the search area and optimized the matching order effectively.The second step was edge point matching.By using the matched corners as datum marks,the edge constraint established in the first step could limit the search area in several pixels.Experimental results show that the proposed algorithm can acquire high correct matching ratio and fast matching speed.

In view of globe lens grinding,multi-component glass substrate ion exchange and crystallization of the glass substrate of light and heat,a method is proposed for designing,manufacturing aplanatic double-convex aspherical waveguide lens and testing its parameter.According to the method based on integrated optical principle,we develop a kind of optical aplanatic double-convex aspherical waveguide lens.It owns large numerical aperture,high magnification,simple structure and good consistency.Furthermore,the lens approaches to diffraction limit and uses light engraving and copying craft.The experimental results that NA= 0.35,β= 6.932 and light distribution corresponding with theoretical results meet the requirements of design.

A kind of single beam Ultraviolet and Vacuum Ultraviolet(UV-VUV) spectrophotometer is designed based on the principle of spectrophotometer.It primarily consists of a 150 W deuterium lamp,a toroidal mirror,a Seya-Namioka holographic concave grating monochromator,a collimator,samples/detector rotators,a photomultiplier tube and a system controlled by computer.The optical system of the spectrophotometer is optimized using ZEMAX optical design software,and a sine bar mechanism for wavelength scanning is designed.The analyzed results show that it satisfies the design requirements of the spectrophotometer.Spectral region from 115 nm to 400 nm can be scanned,and its spectral resolution is less than 0.5 nm with wavelength accuracy less than 0.1 nm.

In order to meet the special requirements of a linear Charge-coupled Device(CCD) camera used in a 2-D localizer system,a new video lens design is proposed,in which a spherical lens is used as the front element and a cylindrical lens is used as the back element.The image plane of the front element and the object plane of the back element share the same position.The video lens has the wide field of view,small distortion,and highly imaging ability.By the lens,the spots equidistant from the basic shaft(eg.x shaft) can be imaged into the same line.The design considerations and the characteristics of aberrations are discussed.A useful design results with f′=-20 mm,2ω=74°,distortion less than 1.6% are provided.The aberration curves and calculated Modulation Transfer Function(MTF) of the video lens are also given.

Aimed at the resisting noise performance of iris localization,a new algorithm for iris slocalization based on iris boundaries recognition was presented.The features of iris inner border and outer border were extracted by image geometric moment function,which was used to recognize the borders of iris by trained Support Vector Machines(SVM) classifier.Finally,the parameters of iris borders were solved by the Hough transform,and the iris was localized.The proposed method was testified on CASIA iris database and the result was satisfying.The data from the simulation experiment show that the proposed method is computationally efficient and has higher correct ratio.Furthermore,the proposed method is less sensitive to obscuring element such as eyelashes and eyelids.

High-precision prealigner is a crucial component in wafer transport device for modern photolithographer.Firstly,design and realization of a prealigner were presented.A three-freedom-degree prealigner mechanism,including rotation stage,centration stage and lift stage,was illustrated.A CCD sensor with liner array was used to detect the edge of the wafer.Least square circle-fitting algorithm was adopted for positioning the center of the wafer and its notch.Effectiveness of the prealignment method was validated by experiments.Then,control system configuration of the wafer transport device was given,and an approach for re-locating,which was used to calculate the wafer eccentricity,was briefly introduced.Finally,experiments of repetitive precision are carried out and the results show that precision of the wafer transport device can be effectively improved based on the prealigner.

Referring to the angular measurement of stabilized platform system with Ring Laser Gyro(RLG),high accuracy,high stability and the appropriate dynamic range are important design requirements.RENISHAW grating angular encoder was the better choices,and the cyclone FPGA chip was selected as signal-processing circuit.Angular encoder signal was synchronized,filtered digitally,doubled frequency and discriminated phase before it was counted.Interface circuit was designed to implement communication between FPGA and the Master-Computer.RLG was applied to calibrate the grating angular measurement.The calibration result shows that by using this set of relative angular measurement devices angular resolution of 3″,measurement accuracy of ±8″ and dynamic rate range of 0 °/s～360 °/s are obtained steadily and accurately.

We demonstrate a simulation method to estimate the bias error in fiber optic gyroscope caused by the even order distortion.The bias error can be expressed by the Bessel functions and trigonometric identities.Corresponding to theoretical simulation,the bias error varies dependence on the even order distortion crosstalk and birefringence phase angle.The bias error can be minimized by choosing the modulation depth around 3π./5 The even order distortion is measured by using the spectrum analyzer since the even order distortion is a main factor in the causing bias error.Moreover,the bias error is measured through the output of fiber optic gyroscope.The experiment data of the bias error is fit well with the simulation result.

Manual adjustment of lighting intensity to achieve high-quality images brought about functional limitations in fully automatic micro-assembly and micro-manipulation.An illumination optimum method based on image analysis in spatial domain or frequency domain was presented.An illumination auto adjustment system was designed and realized to test and solve above issues.Linearity of image average grayscale under the adjustment can reach 1.3%.This paper firstly proposed image quality assessment operators based on high frequency of microscopic images.Illumination intensity can be adjusted automatically by ways of image analysis.Illumination evaluation functions Acon~~Tam and Astd were presented.Experimental results show that the correct illumination configurations are especially important for high resolution microscope and the quality of illumination directly affects the accuracy and repeatability of micromanipulation.

To better understand the characteristics of photonic jets which are generated at the shadow side surfaces of cylindrical or spherical scatterer with their diameters much larger than the wavelength,we used the rigorous Mie theory,with the software Matlab 7.1,to calculate the spatial distribution of the near-field external and internal electromagnetic field of a single spherical scatterer.The numerical errors of the boundary condition are checked and the basic properties of photonic jets outside the spheres in the linearly polarized case,namely the magnitude and location of the intensity peak and the decay length,are discussed.Two distinctive intensity peaks inside the sphere are also analyzed.Besides,the influence of the radius and the refractive index of the sphere on the total field are determined,and explanations are proposed.