For the characters of long-distance flight target and the stringent requirements of time synchronization standard in time-measurement of different locations, we put forward a high-precision time-measurement system based on GPS. The system combines GPS time service with local crystal oscillator clock, which uses GPS second pulse to make time synchronization and local crystal oscillator clock to measure precise time of the start and stop of flight target. Meanwhile, it gives a way to realize high-precision time synchronization from monitoring and compensating GPS second pulse. Experiments show that the system is able to control the deviation less than 300 μs within 400 s’ measurement, and the equivalent precision can reach 7.5×10-7.

The color-based histogram is robust against noise and partial occlusion, but suffers from the presence of the confusing colors in the background. An efficient image sequence tracking method was presented based on multiple cues in Sequential Monte Carlo (SMC). The background-weighted color histogram was combined with edge histogram into SMC for tracking. Color histograms and edge histograms were used to model the object observations likelihoods function. The observations were used to obtain a posterior probability distribution for the location of the object in the sequence images based on SMC. It can be seen from the experiment that the combination of color histogram and edge histogram based on SMC can achieve more robustness and efficient tracking.

To satisfy the need of video moving object locating in intelligent video surveillance scenes, video moving object locating technology based on nonparametric kernel density estimation is proposed. Nonparametric density estimation operation is firstly used on each foreground sample point that stands for video moving objects, and the sample point which has maximum density values is chosen as the first object center. And then other positions of video moving objects are gradually located by modifying the density value of sample point. This object locating method based on kernel density estimation is a further development of subtractive clustering object locating method. Using this method, kernel function could be flexibly chosen to estimate sample point’s density values according to different locating application scenes. Experiment results show that the proposed method has much more flexibility in sample point’s density estimation with satisfying locating results.

Using single point diamond turning method to process KDP crystals, the micro-waviness, which has a significant impact on the laser induced damage threshold, will be left on the machined surface. The effect of micro-waviness on near-field light intensity distribution was analyzed by Finite Difference Time Domain (FDTD) method. Results indicate that: the micro-waviness will distort the incident wave and lead to periodic distribution of strength and weakness of light intensity along x- and z-axis inside the KDP crystal. When the micro-waviness period is below 0.7 μm, it has a low influence on the modulation of light intensity inside the crystal. Meanwhile, the light intensity modulates around the micro-waviness in the form of evanescent wave. Relative light intensity reaches the maximum in the period of 1.064 μm, and remains unchanged above the period of 3 μm. In the common range of the micro-waviness amplitude, the relative light intensity increases linearly as the micro-waviness amplitude increases, with different slope in different micro-waviness period.

The direction of the development for the star tracker is to realize the small size, high-precision, large field of view, multi-parameter, and high reliability. Thus, color temperature, as the secondary gist of the attitude discrimination of the star tracker, has been added into the research. This article focused on the definition of the star associated with the color temperature, and mainly researched achieving a wide range of adjustable, stable and reliable color temperature light source. In order to achieve this goal, based on analyzing the color science, photometric and radiometric studies, we improved the method of calculating the color addition, and built the entire experiment platform based on TMS320F2812 DSC and integrating sphere. Experimental results show that: the color temperature from 2 600 K to 10 000 K continuous variation can be achieved through this method. Under the same conditions (outside ambient temperature), the stability and consistency of color temperature had been validated. Moreover, it is pointed out that the most important reasons affecting the stability of the color temperature is the ambient temperature.

A novel scheme for 0.1 THz optical Tera-Hz signal generation in ToF(THz over Fiber) system is proposed and analyzed. The ToF system used widely in broadband access networks has more stringent requirements on the cost, while the scheme proposed can reduce the system costs. The scheme using frequency quadruplication is based on two Mach-Zender modulators in a certain parameter setting, and we clarify the frequency-multiplication principle from the perspective of both mathematics and physics and the operation principle of Mach-Zender modulator. The condition of the side-band generation and the method for suppressing the band are investigated, and the parameters of suppression ratio are obtained. Furthermore, the tolerance range of the device parameters in a certain suppression ratio is studied, in order to provide reliable basis for the study of device.

In order to replace the commonly used normal diameter Polarization Maintaining (PM) fiber with the ultra-fine PM fiber in fiber optic gyroscopes and other key components, the method of determining its polarization direction needs to be improved and optimized. First, theoretical images of side-view-image method were obtained by computer simulation using ray-tracing method. Then, the situations were verified whose simulated results were better, and the experimental curves were Fourier expanded to suppress random noise. The three-bright-line method whose experimental result was better was improved into three-bright-line-four-feature-point method, and this method was verified by experiments. Finally, the repeatability of this method was measured. Experimental results indicate that the three-bright-line-four-feature-point method has greatly increased the anti-jamming capability of the process of ultra-fine PM fiber polarization direction’s determining, and the repeatability achieves 1.6°. This result has reached the repeatability of common PM fiber polarization direction’s determining, and met the requirement of practical applications..

To rectify the nonlinear relationship between drive voltage and transmission wavelength of the fiber Fabry-Perot tunable filter, a method based on BP neural network was presented. In this method, the drive voltage was input to the BP neural network and the corrected transmission wavelength was got from the output. In order to compare them, linear and quadratic curve fitting method was discussed，too. A set of drive voltage was used as input to the three methods. Compared with the standard value reading from the optical fiber sensor analyzer Si720, the transmission wavelength obtained by the BP neural network has the average error 0.11nm and the maximum error 0.19 nm. In comparison with the other two methods, the average error decreases by 80.6% and 19.7%, and the maximum error decreases by 83.8% and 26.6%, respectively.

The five-point algorithm of essential matrix is a common way to achieve relative orientation of the two-view images in 3D measuring. Polynomial solving techniques, which lead to polysemia while computing, are always adopted during the computing process. In order to determine the right solution, two improved methods for five-point algorithm are proposed to avoid multi-solutions. First of all, the inconsistent solutions of physical model were excluded with cheirality constraint. Secondly, the rest error solutions can be solved by computing sums of Sampson distance of all the common points or re-projection residual. In the two-view images, the minimum value among sums is just the correct orientation parameter values. Both simulation and real images experiments have proved the feasibility and correctness of the two tactics. In most cases, methods based on Sampson are much quicker than that based on re-projection.

Combining electric field distribution of Gaussian beam and plane-wave analysis method, the photo-electric signal analytical expression of homodyne vibrometer system irradiated by Gaussian beam was deduced. And relative measurement error caused by Gaussian beam was analyzed. The relationships between relative measurement error and the distance from detector to laser beam transmission axis, the distance from detector to beam-waist, vibration displacement and beam-waist radius were discussed, respectively. The homodyne vibrometer system of specific parameters was simulated numerically and the simulation results coincided with the theoretical analysis. The results show that the relationship between measurement error and vibration displacement and beam-waist radius is complex. Measurement error increases with the increase of vibration displacement while fluctuating cyclically. There’s a zero point and maxima in the relationship curve of beam-waist radius and measurement error. It provides a theoretical basis for error analysis in high-precision laser measurement.

As an important parameter of the coherent optical system,the linewidth of narrow fiber laser must be measured accurately. Compared with other currently available linewidth measurement techniques, the Delayed Self-heterodyne Interferometer (DSHI) technique is the most promising one. In this paper, theory of the DSHI for determining the linewidth of lasers is described in detail. The MATLAB emulator is programmed to analyze the effects of fiber delay line on the results. Furthermore, an experimental setup based on the DSHI is constructed to measure IPG’s fiber laser linewidth. We substitute oscilloscope for the spectrum analyzer in this system, and the software algorithm is designed for the analysis of the photocurrent signal. From numerical analysis, the measured linewidth of the IPG’s fiber laser is about 16kHz, which is in the range of an ideal accuracy.

Since non-contacting displacement sensors with high precision have been used widely, this paper presents one of examples, which is the spectral confocal displacement sensor. Firstly, the working principle of the sensor is described, and then the design and evaluation schema of the system is proposed. Finally, we design a specific case of the spectral confocal displacement sensor objective lens based on CODE V macro language and MATLAB. The sensor whose work-distance is 0~1.173 mm has the measuring accuracy better than 5 μm. The lens is combined with a single lens and doublet lens. In the work wavelength, each single-wavelength has the corresponding largest RMS radius of 2.8 μm. Through linear regression fitting of the defocusing amount and linearity of the wavelength, it is obtained that the coefficient of determination is 0.985 9. So the linearity of measurement is well.

A novel geometric robust zero-watermarking scheme, which uses Computer Generated Holograms (CGHs) as the watermarks, is presented. In the zero-watermark construction process, the host image is decomposed with discrete wavelet transform and the low frequency sub-band is divided into nonoverlapping blocks. Singular Value Decomposition (SVD) is applied to every block. A quantization matrix is obtained by quantizing the largest singular value of each block. Using the quantization matrix and the mark CGH, the zero-watermark which is registered in the database of IPR can be constructed. In the detection process, the geometric distortion parameters of the attacked image are estimated first by the proposed improved geometric correction method based on the Zernike moments and invariant centroid. Then, the mark hologram is extracted from the recovered image. Compared with traditional zero-watermarking methods and digital hologram watermarking methods, the suggested watermarking scheme provides better performance for resisting geometric attacks. The experimental results demonstrate that the proposed watermarking scheme is robust against rotation, scaling, translation, flipping and combined attack, as well as a variety of common manipulations such as filtering, additive noise, JPEG compression, blurring, cropping and contrast enhancement.

According to the redundant and complementary information among polarization parameter images, a polarization imagery fusion method based on human visual characteristic was proposed. Firstly, by fusion of the images of linear degree of polarization and angle of polarization using energy feature, the polarization feature image was obtained. And then, quincunx sampling lifting scheme wavelet transform was used to fuse the images of polarization feature and total intensity. Finally, based on the feature that human vision is sensitive to the local contrast, the image of linear degree of polarization was adopted to enhance the fused result. Experimental results have shown that the fused images not only make the polarization information of the scene prominent, but also contain abundant intensity information, as well as the contrast between the target and the background is remarkably improved.

An approach for estimating object pose from coplanar circles and implementation 3D registration is proposed in augmented reality. Provided that intrinsic parameters of camera has been calibrated, according to common virtual intersecting points between absolute conic and coplanar circles，rotation component of pose is estimated through calculating the equation of vanishing line, determining the centers of circles as well as axes direction of frame. Furthermore, translation component is finally calculated so that object pose is determined completely. The algorithm requires fewer preconditions with low computational cost, which is more readily satisfied in industrial scenario. Experiments on synthetic datum and real object show that the algorithm is correct and robust to noise and occlusion.

Image restoration of sparse aperture imaging system is an important step of sparse aperture imaging technology. Exploratory research on restoration of blurred image caused by misaligned sparse aperture imaging system is carried out by using phase diversity based on defocus, and adaptive genetic algorithm is adopted as numerical method. At the same time, the restoration of image added Gaussian noise is also researched. The result of simulation experiment shows that phase diversity is a very effective method for restoration of sparse aperture system image and retrieval of sparse aperture system aberration.

Since the result of evaluation usually had ambiguity, it is difficult to define the performance of image matching algorithm with “good” or “bad”. So the performance of image matching algorithm can be evaluated by comprehensive evaluation method in fuzzy theories, which indicated the degree of evaluation results belonging to a certain review. Finally, it can get more scientific evaluation results. Combined with image matching algorithm performance evaluation of multiple criteria, the modeling course and the method for determining the weight were explained. With the evaluation experiments, the performance of gray correlation matching algorithm and phase correlation matching algorithm was evaluated by using the model. Compared with the single criteria evaluation architecture, the evaluation result is better. The practicality and effectiveness of the evaluation model are proved.

Fingerprint enhancement is an essential preprocessing step and it is crucial for the efficiency of the fingerprint recognition algorithm. An algorithm based on curvelet domain is proposed. The input image is decomposed into coarse and fine scales coefficients, and directional filters and soft threshold function are applied fully to enhance image and reduce noise. In curvelet domain, directional filters are used to enhance coarse scales coefficients. Meanwhile, soft threshold function is used to reduce noise of fine scales coefficients, which is much different from the conventional methods based on spatial domain and frequency domain. Experiments are carried out on FVC2004 databases. For fingerprint with bad quality, the results indicate that the proposed algorithm can better enhance fingerprint images and reduce noise than traditional methods, and need less time.

Because of the limitation of single biometric recognition, a new multimodal biometric system based on hand vein, iris and fingerprint was proposed. Firstly, the hand vein image, iris image and fingerprint image were separately sent to image preprocessing module, feature extraction module and feature matching module, respectively. And the respective matching scores were obtained. The effects of score normalization on recognition performance were analyzed. Tanh normalization method was employed to perform the three scores normalization. Finally, the weighted sum principle was used to fulfill the scores fusion. Minimum distance classifier was adopted to personal identity decision. The experiments results show that Equal Error Rate (EER) of proposed recognition algorithm is 0.009%. When the False Acceptance Rate (FAR) is close to 0, the corresponding False Rejection Rate (FRR) at the same threshold is 0.2%.

Finger' decrustation, ageing and the image error caused by living fingerprint sensor may affect the imaging quality of fingerprint easily. In order to solve the problems which exist in the inferior quality fingerprint image, such as line breaks, the adhesion of ridge and valley as well as the regional blur caused by finger decrustation, a method based on the analysis of fingerprint image energy is proposed. The algorithm has analyzed and calculated the intensity field, gradient field, directions field and the frequency field of fingerprint image, and then unified the concrete information as well as the principle of Gabor filter to enhance the fingerprint image progressively. The practice indicates that the whole set of algorithm has better enhanced performance and can improve the image quality distinctly.

The chemical polishing techniques was introduced to fabricate Continuous Phase Plate (CPP). The basic theory of Marangoni interface effect and the removal function were discussed briefly. According to the characteristics of the CPP’s surface, the chemical grinding wheel was designed. The technics experiments have been done to find the optimal effect of NH4F addition on etching solution and the effect of the temperature fluctuation. While the concentration of NH4F is 10%~13%, the best etching rate can be gotten. Combined with the process parameters, the process flow is designed and some typical CPP fabrication results are achieved.

The method of spatial convolution has been applied to analyze the imaging process of CCD imaging system, which involves optical imaging process and CCD integral sampling process. The way of changing the ratio between Ground Resolution Distance (GRD) and Ground Sampled Distance (GSD) (Kell coefficient) to improve the image resolution was introduced. Imaging process of the target was simulated with different Kell coefficient values. The method of comparing the mean modulation values in a period of the images on CCD which was regarded as the evaluation criteria was proposed to evaluate the image quality of the CCD aerial surveying camera. The results show that when Kell coefficient equals 2.8, the target characteristics can be distinguished.

In order to improve water quality retrievals of multi-spectral image accurately, a model for water quality remote retrieval is put forward based on Support Vector Regression (SVR) with parameters optimized by genetic algorithm. The model based on high-resolution multi-spectral remote SPOT-5 data and the water quality field data, uses CV to estimate the promote error. And parameters of SVR model are optimized by Genetic Algorithm. The global optimization of model parameters is achieved. Then, water quality is retrieved by the trained SVR. The proposed model is applied to the water quality retrievals of Weihe River in Shaanxi Province. The result of experiment shows that the developed model has more accuracy than that of the routine linear regression model，which provides a new approach for remote sensing monitoring of environment in inland rivers.

In order to satisfy the modern power industry demands of high voltage and large power, a new type of optical voltage transducer was presented. The structure of optical voltage transducer was designed with converse piezoelectric effect of quartz crystal and interference between modes in polarization maintenance fiber. The piezoelectric deformation of the cylinder-shaped transducer crystal induced by an applied AC voltage was sensed by an elliptical-core dual-mode fiber wound on the circumferential surface. White-light interferometry and active homodyne phase tracking detect the resulting modulation of the differential phase between LP01 mode and LP11 mode in fiber. According to the relation between group modal refractive index difference and normalized frequency, the optical fiber's parameter is researched. The feasibility of the designed voltage transducer is verified with experiments. The results show that the voltage transducer based on the principle mentioned above possesses better accuracy and linearity, which can provide theory basis for optical high voltage transducer.

As a modified approach to improve the properties of the classical Morphological Correlation (MC)，a Laplace Filtering-based Morphological Correlator (LBMC) is proposed for pattern recognition. Different from classical MC, for the LBMC, the Joint Power Spectrum (JPS) summation of the MC is filtered by power spectrum of Laplace operator before final inverse Fourier transform. Computer simulation results show that, compared with the Linear Correlation (LC) and the conventional MC, LBMC provides better discrimination capability with sharp and strong unmistakable correlation signal, and its ACR (autocorrelation peak intensity to cross correlation peak intensity ratio) and Peak to Noise Ratio (PNR) are more stable when input scene is corrupted by outlier noise (salt-and-pepper noise). Although the LBMC is not illumination invariant when multiplication illumination factor is larger than unity, considerable discrimination capability is still obtained.

The slope of the curve optical power VS. voltage at ATT=1 dB is required to be less than 1 dB/V, which is not met by most of VOA. The traditional method is to control the gain of multi-stage operational amplifiers by the output feedback. However, the discontinuity points on gain curve are inevitable. Based on the pure non-linear circuit, the requirement can be met. Besides flattening the slope @ATT=1 dB at specific points, the curve still keeps the original value and there is not any discontinuous points. The theory is analyzed, the circuit is simulated by Cadence Orcad Pspice software，and the conditions of zero slope the derivative of the curve is zero) are derived. The diversity of JFET and MATT VOA is analyzed and the temperature characteristics are investigated. Based on the technique above, the circuit occupies small space and its cost is low, 8 pcs MATT VOA modules are made successfully and the optical and electrical test results are shown.