For a specified digital optoelectronic tracking system, we analyze a kind of equalized clearance nonlinearity generated by the quantification and optoelectronic sensor delay when the system is in TV tracking mode. Two main theoretical expressions of nonlinearity are shown. Then, the measured self-oscillation is characterized with the description function. The double mode PWM driver is used in the digital optoelectronic tracking system. In positioning experiments, the maximum peak-peak value of positioning error is less than 22.23″. The results show that zero-jitter feature of the double mode PWM driver can greatly restrain the self-oscillation nonlinearity.

Servo control is one of the key technologies of the photoelectric stabilized platform. Improving the control rigidity may strengthen the isolation effect, but may result in system instability at the same time. Based on the theory and experiments, the paper builds up the simplified mathematical model of the servo control system, deduces the system control rigidity expression, and points out the method to improve control rigidity. According to Routh stability criterion, the analytic stability condition is solved, which shows the conflict between the stability and control rigidity and that gyroscope is the key factor of the conflict. The results are validated by simulations and experiments, which may provide the reference for engineering application.

Chord transformation and the description of chord space feature are studied in detail, and the concrete approach for extracting geometric feature is achieved experimentally. Meanwhile，the robustness of this feature is proved through experiment. The feature has invariant character for translation, rotation and scale, and some degree of tolerance for affine transformation. On the other side, after filtering original image with the feature of a specifically shape of model in the chord space by inverse chord transformation, the result holds the edges of objects having the same or similar shape with the model while restraining other edges. Experimental results show that it is effective to use the feature for the similarity match between (geometric) shapes. Moreover, the approach is easy to be understood which can be implemented in various fields such as object recognition, target tracking, etc.

A novel Multi-sensor Order Statistic Unscented Probabilistic Data Association Algorithm (MSOSUPDA) is proposed for the multi-sensor multi-target tracking problem of nonlinear system in clutter. In the new algorithm, the problem of interest is first translated into multiple nonlinear single-sensor multi-target tracking problems. Then, the association of measurements of single sensor to multiple tracks is implemented according to the principle of Order Statistics Probabilistic Data Association (OSPDA). Based on these, Unscented Kalman Filter (UKF) is used for the propagation of state distribution in nonlinear system and the MSOSUPDA is derived. Compared with the MSJPDA/EKF, the accuracy and robustness of MSOSUPDA are improved. Furthermore, computation complexity of the proposed algorithm is decreased obviously on account of the use of OSPDA. According to the simulation results, the divergence ratio and the processing time of our proposed algorithm are equal to 19 and 70 percent of the MSJPDA/EKF algorithm respectively.

Aiming at the designed spinning prism Q-switch, the refractive index of the prism was considered for the first time in the optical acceleration, and the influence of the index of the prism was calculated in detail based on the theory of fourfold optical acceleration. The calculated results show that compared with the flat mirror, the rotating speed of the prism will “diminish” when the refractive index is considered, which means that the smaller refractive index of the prism have better optical acceleration. In one word, the refractive index of the prism has negative influence on the optical acceleration. Therefore, in the design of the optical acceleration device, the refractive index of the prism should be considered carefully in order to obtain the best optical acceleration.

DNA-CTMA complexes were prepared by ion-change method and DNA-CTMA thin films were fabricated by self-assembled and spin-coating methods. The photo-electrical properties of DNA-CTMA thin films were characterized by the measurement of the absorption spectra in the UV/vis/inf and far-infrared, refractive index and Alternating Current (AC) resistivity. The experiment results show that the transmissivity of the DNA-CTMA thin films is high in the telecommunication wavelengths, and there is a strong absorption only in the ultraviolet range of 230~315 nm and the infrared range of 2.7~3.34 μm, including two strong absorption peaks at 3 418 nm and 3 504 nm in the absorption spectra. Furthermore, in the preparation process of the films, the some factors of influencing the refractive index of film were analyzed, such as the butanol remained in film, temperature and the dry time. And the AC resistivity of 50 bp short DNA-CTMA thin film is less 2~3 orders of magnitude than the 2 000 bp longer DNA-CTMA thin film.

An air leakage localization method based on infrared image fusion was investigated. Modeling and simulation of heat transfer coupled with steady air-leak flow were addressed by the finite element method. The temperature difference in leak throttling process was derived from the expression of Joule-Thomson coefficient. In the experiments, metallic and metalloid test objects were charged with cryogenic compressed air respectively, which enhanced the phenomenon, and their temperature fields were acquired by an infrared camera. Leak diagnostic criterion was set as different values referring to the thermal conductivities of various tested materials, and entropy fusion operation among the infrared images captured before and after inflation was conducted. Then, the leak positions were located by a certain strategy. The advantages of the method were demonstrated by experiment, especially in eliminating the interference of surface defects and the influence of shapes by employing the entropy fusion algorithm.

Finding gradient maximum and local maximum based on multi-scale Canny operator edge detection is really the optimization for two dimension multi-element function. The maximal function gradient or local maximum which was derived by the traditional analytics is approximate and local. A new multi-scale edge detection algorithm is proposed by a genetic optimum searching algorithm. To upgrade the genetic algorithm convergence about edge detection, an improved GA+SA+TABU is used in order to overcome the defects of local searching in the general genetic algorithm and upgrade the whole resolution. Alternating optimization tactics are utilized by combining the general algorithm and heuristic searching methods. The experimental results show that the proposed algorithm applied to IR target image segmentation could result in copious details, single edge and exact location.

One of the difficulties in light-probe 3D vision measurement is the stereo matching of the light-spot image points in the left and right images. Therefore, the SoftPoist algorithm was adopted to transform the difficulty to a position and orientation estimation problem under the condition that the correspondences of the 2D image points and the 3D object points are unknown. By combining the correspondence problem and the position and orientation estimation problem together to perform bundle optimization, the matching of the light-spot image points and the spatial light-emitting-spots of the light-probe could be completed, respectively, in order to achieve the stereo matching of the light-spot image points in the left and right images. The proposed method is based on model and is not related to the spatial point distribution, point number and the image gray level. The experiment results show that the proposed method is feasible and can be also used for feature points recognition.

In order to realize rapid measurement on blade of large water turbine, a new approach based on photogrammetry was proposed and implemented. The working procedure of the new method is divided into several phases: First, paste reference points on the surface of the object, and then, take a group of pictures from different view as 2D original data. Once these pictures are loaded on the measurement software XJTUDP, it begins to inspect each picture, detect and locate the reference points. Finally, according to their 2D coordinates on different pictures, 3D coordinates of the reference points are reconstructed, which in fact are the different projection of the same reference points. As a result, a great deal of 3D reference points form the model of the object. Consequently, further measurement about the exterior dimension could be carried out. Experiment result shows that, compared with traditional methods, the new method is timesaving, low cost and can be applied on-line measurement easily. Furthermore, it is immunity from dust, vibration, humidity and temperature variety.

A novel method is presented based on camera self-calibration and quasi-dense match for environmental terrain reconstruction of planetary rover. The terrain map surrounding planetary rover is generated by selecting Scale Invariant Feature Transform (SIFT) point, camera self-calibration using Kruppa equations, quasi-dense match and 3D reconstruction. Firstly, camera parameters can be obtained by camera self-calibration using Kruppa equations without geometrical structure presenting on special scene of the planetary topography and geomorphology. Then, based on SIFT features point, quasi-dense match can achieve more reliable and accurate matching points. Finally, According to the computed results, dense and high precision terrain map of planetary rover can be obtained quickly. The experimental results with simulated environment show the method is effective.

In order to solve the problem that the border, between dark crescent and bright crescent, is hard to be detected in the pupil images of infrared eccentric photorefraction, a geometrical plane parameter measurement method based on pupil images processing is proposed in this paper. The key technologies are curvelet transform and least-squares fitting. Firstly, we obtained the pupil images by the infrared eccentric photography system. Secondly, curvelet was utilized to reduce the noise from the images, and the rough edges could be located. Thirdly, by least-squares fitting to the edge, we could acquire the geometrical plane parameters of pupil images more accurately. Finally, pupil diopter could be calculated. The experimental results demonstrate the accuracy, stability and validity of the measurement system.

A novel full-duplex Radio-over-fiber (ROF) system using a single light source at Central Station (CS) is proposed and demonstrated. Based on Cross-gain Modulation (XGM) and optical filtering technique, the scheme is employed to generate 40 GHz optical millimeter-wave at CS for downlink transmission by using a Semiconductor Optical Amplifier (SOA). Baseband signal of the down-link data is only modulated on one of the optical carrier sidebands while the up-link data will be modulated on optical carrier in the base station and sent back to the CS. The bidirectional full-duplex 2.5 Gbit/s data is successfully transmitted over 40 km standard Single-mode Fiber (SMF). The upstream channel is less than 0.5 dBm power penalty and the downstream channel is 3.5 dBm power penalty. This system shows good performance over long-distance delivery and has important applicable value in high radio frequency sector and multi-channel full-duplex system.

The FSO-OFDM (Free Space Optical Communication-Orthogonal Frequency Division Multiplexing) model based on light polarization is designed. We expatiated on the principles of the transceivers and the scatter channel. As the scatter channel is a frequency selective fading channel, each subchannel impulse response keeps correlation during coherent time. The correlation of the same subchannel impulse response is utilized during seriate OFDM symbols and Linear Mean Squared (LMS) algorithm is adopted to estimate the iterative initial value of the future symbols, then to get the channel impulse response. The horizontal and vertical polarization light paths are the same, so the characteristics of atmosphere channel are the same. At last, the more accurate channel impulse response is gotten after weighting the two estimative channel impulse responses. The simulation results indicate that the LMS algorithm can track the channel and get 2dB improvement of system BER.

By comparing the cross-gain effect of Semiconductor Optical Amplifier (SOA), the cross-MQW-EAM absorption wavelength conversion principle is studied based on the simple structure and high efficiency of EAM wavelength conversion, and experimental devices of wavelength conversion is designed. Furthermore, the existing cross-absorption effect model is modified so that it can be used for the situation of multi-beam signals injection. Finally, the EAM mathematical model is analyzed in two conditions. The carrier concentration relation graph in single-Gaussian pulse injection and strong signal injection could be obtained by model simulation. The simulation results are consistent with the fact.

A texture image retrieval approach based on Dual-tree Contourlet (DT-Contourlet) transform is proposed. DT-Contourlet is constructed by the pyramidal dual-tree directional filter bank. It maintains the flexibility direction selectivity of the contourlet transform, and overcomes the limitation of traditional contourlet which namely lacks of shift invariance．According to the statistical characteristics of DT-Contourlet coefficients, each subband is modeled by generalized Gaussian distribution, and the moment estimator is used to extract model parameters. Finally, the texture image retrieval is implemented by computing the Kullback–Leibler Distance (KLD) between estimated models. To evaluate the retrieval performance, texture database including 800 texture images from Bradatz album is created. Experimental results indicate that, compared with traditional approaches based on wavelet and contourlet, the proposed approach improves average retrieval rate with 10.46% and 6.94% respectively and retains high computational efficiency.

Since the convergence velocity of mean shift is too slow, fast mean shift for image segmentation is proposed based on conjugate gradient. Conjugate gradient method is characterized by simple, low memory requirements and local and global convergence properties. Moreover, the convergence velocity of conjugate gradient method is between steepest descent method and Newton method. The new algorithm makes use of the properties of conjugate gradient method to improve the convergence velocity of traditional mean shift by interleaved execution of mean shift and conjugate gradient method. Experimental results on synthetic and real images show that new algorithm not only improves the velocity of classical mean shift, but also keeps better segmented result in image segmentation.

A new and simple edge detection method is proposed. Combined with the intuitive appearance of the gradient on the edge, the statistics property of the histogram and the local processing ability of block-based operation, the Blocked Gradient Histogram (BGH) is generated. Then, mutual recognition about the peaks of BGH between neighboring blocks is implemented so that the edge is not sensitive to the gradient intensity. Accordingly, more edges are detected. Results and performance of the proposed method are better than that of other involved ones. Moreover, it is not sensitive to the noise.

During the process of deblurring image, the ring artifacts are inevitable, which seriously affects the definition of the deblurred image. If the blurring kernel is inaccurate, the ring artifacts will be more serious. By analyzing the gradient distribution histogram of the images during the deblurring process, we demonstrated the necessity of adding the nature image gradient constraint to the image restoration. A variable substitution scheme was established to simplify the computation. According to the calculated data, the impact of ring artifacts is not serious in the synthetic image or the real motion blur image by this approach. And the image gradient distribution of calculation result was closer to the latent image than the RL. Moreover, we cloud get a good result even if the motion deblurring kernel was inaccurate.

An efficient parallel processing technique and a scalable architecture of lifting Discrete Wavelet Transform (DWT) were proposed for high-speed implementation. The independence of the same stage of lifting operations of the multiple input data for lifting DWT was exploited to increase the amount of parallelism, which efficiently increases the data processing capacity of system. New high throughput architecture of 1-D (5, 3) DWT was introduced especially. The throughput of the proposed design can be flexibly extended to 2J-samples/clock-cycle, which provides a new alternative for high-speed implementation of DWT. Theoretical analysis and experimental results show the effectiveness of the proposed design.

A new kind of watermark-embedding and detecting algorithm based on stationary wavelet transform was proposed. Firstly, the digital watermarking was transformed randomly (Arnold transformation), and then encrypted by chaos. The encrypted watermarking was transformed to one-dimensional row vector, and the pixel value was sorted. The coefficient of primitive image of stationary wavelet transformation was expanded to one-dimensional row vector, too, and then the sorted watermarking was embedded to the sorted low frequency and turned to two dimensions. Secondly, the image is reconstructed by coefficients of high-frequency. The process of watermarking extracted is the inversion of the process of watermarking embedded. Finally, the watermarking was withdrawn and the primitive watermarking was obtained after the anti-Arnold transformation. This algorithm chose the appropriate position to embed the watermarking. Experimental results indicate that the algorithm not only enables the watermarking to have the very good invisibility, but also makes the watermarking have very strong robustness to the general image attacks, such as noise, filter, rotation, compression and so on.

Concerning about the Temporal Information (TI) detection for video quality assessment, a new method is proposed based on the definition and algorithm recommended by the ANSI T1.801.03-1996 and ITU-T Recommendation P.910. At first, image of the temporal derivative would be changed into the histogram domain, and then the statistic calculation (RMS and so on) was done for the temporal information. The drawback of the conventional method was overcome whose workload would increase greatly with the increase of the number of the pixel in scene image. Experimental results show that the computation complexity is dependent on the gray level of histogram, but is independent of the number of the pixel by the new method. In the same experiment condition, the computational efficiency is improved three times while the performance is similar to the conventional method. The algorithm is so simple and feasible that it could be used for the application of video quality assessment.

Aiming at shift and rotation in fingerprint images, a novel dispersion degree and Ellipsoidal Basis Function Neural Network (EBFNN)-based fingerprint classification algorithm was proposed in this paper. Firstly, feature space was obtained through wavelet transform on fingerprint image. Then, the optimal feature combinations of different dimension were acquired by searching features in the feature space. And the feature vector was determined by studying the changes of divergence degree of those optimal feature combinations along with the dimensions. Finally, EBFNN was trained by the feature vector and fingerprint classification was accomplished. The experimental results on FVC2000 and FVC2002-DB1 show that the average classification accuracy is 91.45% if the number of the hidden neurons is 11. Moreover, the proposed algorithm is robust to shift and rotation in fingerprint images, thus it has some values in practice.

An Extend Kalman Filter (EKF) based on tracking data process was introduced for a multi-sensor laboratory Helmet-mounted Display (HMD) system. Moreover, a set of image preprocessing program was designed to represent head position based on vision. Fast angle compensation procedure (based on translation to rotate pre-rendering) was used to compensate rendering delay dynamically. EKF combined with prediction technique reduces prediction error of measurement and process. Fast angle compensation allows extending the rendering lacking time tolerance of display processing unit. 3-DOF rotating platform verification results show that compared with the original equipment, the filter reduces the noise and delay greatly and thus guarantees the quality of helmet usages.

To improve the precision of the encoder system, a method for error compensation in optical encoder was presented by applying Radial Basis Function (RBF). At first, how to integrate RBF model with error detecting system of encoder was introduced. The simulation analysis can be completed while run error detecting. At the same time, necessary model parameters could be obtained. Then, the parameters were written to DSP to run real error compensation. The optimization for RBF algorithms was introduced to resolve the problems existed in the RBF model, i.e., large amount of data, massive matrix operation and complicated arithmetic, ect. The practical application proves that the simulation effect is direct and clear by combing the model building with error detecting. The test results show that the optimized compensation model meets the requirements of running time without decreasing compensation precision obviously and the precision of system error is improved 1~3 times.

Optimal design of ring resonator in silica optical waveguide for Integration Optical Gyroscope (IOG) is presented by using Wide Angle Beam Propagation Method (WA-BPM). Furthermore, relationship among the finesse of ring resonator, the total loss and the power coupling ratio is analyzed. On the base of optimal design, we fabricated SiO2 optical waveguide ring resonator mixed Ge in Plasma Enhanced Chemical Vapor Deposition (PECVD) method. By testing at λ=1 550 nm, propagation loss and total loss of ring resonator are 0.02 dB/cm and 0.1 dB/circuit, respectively, which can offer academic reference for the integration optical gyroscope’s micromation and high precision.

According to three-wavelength-division-multiplexer, a three-port all-fiber optical Wavelength-divisionmultiplexer was developed by using a new configuration of cascading three 3×3 directional fiber couplers, which was made of three 3×3 single-mode fiber couplers connected by 3 single-mode fibers. Based on the principle of characteristics of fiber transmissivity, the complex amplitude output spectral expression was established and described by using the matrix transfer function. Through numerical simulation and experiment, the influence of the coupling coefficient of the couplers and the lengths of these 3 single-mode fiber on the three-wavelength-division-multiplexer was analyzed. The analytical results indicate that when the lengths of these 3 single-mode fiber are L, L+ΔL, and L+ 2ΔL, respectively and the coupling coefficient of the couplers is some certain values, three-wavelength-division-multiplexer acts as a 3 channel wavelength-division-multiplexer and the port output curves are quite ideal. The experimental sample is tested at 1.55 μm working wavelength and the peak output power interval of 0.8 nm. The experiment results are in good agreement with the analytical ones.