Referring to the possible damage to the traditional flexure gyroscope under the fire impact in a tank based high precision gyro-stabilized electric-optical tracking platform, a fiber rate gyro feedback scheme is adopted. In order to eliminate the steady-state angle error which occurs in rate gyro feedback and traditional PID control scheme, a pole placement scheme is presented to form a PII2 (proportion + integrator + double integrator) controller with pre-filter. Dynamic angle error is 30″ when the platform is under 3 000 gcm torque disturbance and adjustment time is 0.2 s. The steady-state angle error is zero. The results show that the PII2 controller form can improve system stabilization stiffness and reduce steady-state error.

Based on the three close-loop high precision position servo system, the error source is introduced and categorized into two errors. The one can be adjusted by the control algorithm and the other can not be adjusted by the control algorithm. The rounding error of computation, the measurement error of the sensor, the error of the sampling frequency, and the error from the bandwidth are analyzed quantitatively. The transmission and superposition of the error among the loops are analyzed according to the orders from the inner loop to the outer loop at the point of view of the physics kinematics. Through the quantitative analysis of the errors, the proportion of each error in the whole error is shown, which will provide important theoretic basis for the system design. The experiment is carried out and the theory of the error analysis is proved to be feasible.

Aiming to the shortage of conventional friction compensation methods and nonlinear characteristic of friction disturbance, the accuracy of the friction description by the LuGre model is used to analyze the design process of Luenberger-like Observer and Time Delay Estimator (TDE). Then, the method of synthesis of Luenberger-like Observer and TDE is given out as a friction compensation scheme. The experimental results show that, as the friction state is observed by Luenberger-like Observer,the friction disturbance estimator with LuGre friction dynamic model can be effective to friction compensation, and the application limitation that the disturbance is asymptotic variety is solved. At the same time, the survival error of friction compensation and exoteric disturbance are eliminated by TDE.

To delete false intersection results produced by multiple trajectory cameras when they measure the multiple near ground coplanar targets with basic intersection method, a method for deleting false targets and improving the intersection accuracy with partial least squares multiple factors regression analysis is presented. Based on the distance between the multiple intersection results, the false targets can be recognized. The best estimation data of target on earth coordinates can be computed with the Partial Least Squares (PLS) multiple factors regression analysis method. Actual test results show that the method can recognize more than 20 coplanar targets which are near ground, and compared with two station intersection methods, PLS reduces measurement error more than 6″.

By simulating wavefront reconstruction with Zernike polynomials, the performance of the sector lens array configuration in H-S wavefront sensor was tested. Compared with square lens array configuration in H-S wavefront sensor, the sector lens array configuration in H-S wavefront sensor has better performance when the number of the array lens of the two configurations is commensurate. However, the simulating of the wavefront which was measured by the interferometer shows the two H-S wavefront sensors comparatively reconstructed the original wavefront.

The basic concept and procedure of digital in-line holography are shown mainly for particle diameter measurement. The result of the reconstructed image and the particle diameter measurement are affected by some recording parameters, i.e. recording distance, particle density, the size of CCD chip and the pixel size, etc. These recording parameters may even determine the success or failure of the measurement. Two parameters, recording distance and particle density, are mainly discussed for particle diameter measurement through computer simulation. The range of recording distance is given by formula. Simultaneously, a kind of method for particle diameter measurement based on gray gradient by extracting particle edge is put forward and verified by experiment.

The model field coupling theory was applied to calculate the coupling efficiency of a Gaussian laser beam to an optical fiber via a single lens. The dependences of the coupling efficiency on the focal length of the lens were investigated for both single-mode and multi-mode fibers. An experiment was conducted to couple a 532 nm laser beam with a M2 factor of 1.05 to both a single-mode fiber with 3.5 μm-core and a multi-mode fiber with 25 μm-core. Good agreement is obtained between the experimental results and the theoretical calculation.

The phase error for the velocity measurement was analyzed by means of displacement interferometer in high speed diffuse surface, including the error source, its range and its relationship with the system parameters. The expression for the optical field distribution for the optical signal was calculated. The result suggests that the sources of the phase error include two parts: the diffraction effect and the random phase modulation by the diffuse surface. The error range for the displacement and velocity from the phase error is 10-6. Furthermore, we measured the velocity profile for the copper plates under explosive loading by the method. Compared with results from VISAR, experimental result indicates that the fiber displacement interferometer can accomplish the same order of precision with the VISAR.

To overcome the shortcomings of traditional methods, a method of underwater image segmentation based on the discrete fractional Brownian random field was proposed to dispose underwater images. At first, a window was set up, and the centre of window was located at the position of each pixel in the image. The average of fractal dimension in the window was calculated, and it was considered as the fractal feature of the pixel at the centre of window. At last, a threshold was determined according to the graph of fractal dimension, and the segmentation of underwater image was completed. By the normalization of the average absolute intensity difference on surfaces at difference scales, the number of data items used to represent the average absolute intensity difference on surfaces at difference scales was reduced, and the segmentation efficiency was improved. Finally, the results on some typical images were presented. Compared with the results obtained by the segmentation methods of Ostu and Maximum Entropy, it shows that the presented method is robust and efficient in underwater image segmentation.

To achieve real-time 3-D medical image despeckling, a novel fast algorithm and high-performance Very Large Scale Integrated (VLSI) architecture for 3-D median filter were presented. According to the data’s distributing character of the 2-D array ordered by a 2-D partial sorting technique, the proposed method could efficiently reduce the implementation complexity of the 3-D median filter by using a 3-D partial sorting technique. Theoretical analysis and experimental results demonstrate that the presented method could reduce the computation complexity and/or hardware resources by about 33%, which could significantly decrease the computation delay of the 3-D median filter implemented by software, and/or the hardware cost and power consumption.

A new algorithm was proposed for extracting texture in complex multi-texture images. Firstly, polar coordinates method was used to analyze wedge and ring characters of texture frequency spectrum, and the period and directional parameters of texture distribution were computed. Secondly, based on these characters, ring and wedge Gauss pass-band filter banks were constructed to filter the frequency spectrum of multi-texture image. Thirdly, components of corresponding texture were acquired by transforming the filtered frequency spectrum image to time domain image. Lastly, through the rectification of texture area and corresponding post-processing, the true texture was extracted. The experimental results show that each texture is extracted truly and the elemental character is preserved exactly, and even in the area where distribution of texture is non-uniform, the skeleton of texture is still extracted truly.

X3D auto-stereoscopic display needs 8 views as its input video signals so that the user can watch stereoscopic scenes from different angles within a certain range. An auto-stereoscopic display oriented stereoscopic video coding method is proposed, in which the disparity map is generated and compressed at the encoder to lower down the requirement to the user side. By using the decoded disparity map, 6 intermediate views can be easily and quickly rendered at the user side, and the stereoscopic video can be displayed on auto-stereoscopic terminal with multi angles of view. Global filtering and non-edge region filtering are employed to smooth the disparity map. Experimental results show that compared with un-smoothed disparity map, the rate-distortion performance can be improved by the proposed scheme with smoothed disparity map.

The present pose estimation methods are mainly based on geometry analysis or linear transformation, which are complex and are not universal. A new method is proposed based on nonlinear transformation. According to manifold learning theory, different head poses lie on some low dimensional manifolds. Kernel Principal Component Analysis (KPCA) is a nonlinear dimension reduction method. The hidden manifold in the high dimensional space can be successfully embedded to a low dimensional space using KPCA. A pose curve is gotten using KPCA train samples and new pose image is projected onto this curve. The pose angle can be estimated using interpolation method. The disadvantage of traditional linear method is conquered by KPCA and the experimental result shows that the method is effective to estimate head poses. The method to improve the estimating result is suggested based on the experiments.

Referring to the complexity of the eyelid edge structure, many interferential factors and the slowness of the localization speed, large window smoothing filtering, edge points detection in special directions and high threshold edge point screening were used to reduce the interference from the eyelash, eye texture and other ridge-like edge, and the point set was acquired on the edge of the single-layer or the double-layer eyelid rapidly. And then, the curve fitting method was used to process the detected point set to acquire the continuous eyelid edge curve. The first eyelid location was completed. Finally, according to the characters that the inner and outer edge of double-layer eyelid are approximate parallel and the gray distribution changes sharply on the edge of inner eyelid, take the fitting curve acquired in first eyelid location as a template, and shift it down parallel to detect the position where gray value changes abruptly. Then the second time eyelid location was finished. The experiment shows that the location speed can be improved effectively on the basis of ensuring an high location accurate rate.

The algorithm for Iris recognition based on Gabor transform is difficult to choose Gabor parameters. A method for parameters selection in Gabor filter which used in iris recognition is proposed. In this method, the position parameters can be selected according to partitions of image, the size parameters can be chosen by the curve of hamming distance; and the frequency parameters can be gotten based on the proportion to size parameters. With the Gabor filters designed by the method to iris feature extraction, experimental results show that it can extract effectively the characteristics of iris texture for iris recognition and has high accuracy.

The fundamental theory of Dynamic Light Scattering (DLS) was introduced. Several attempted experiments were done to measure the diameter of smoke particles in gas using DLS measurement. By means of a measurement platform containing BI-9000 digital autocorrelation instrument, which was produced by Brookhaven Instrument Corporation, we explored some possibility ways of measuring smoke particles in gas. Several important issues were analyzed in the process of measuring, e.g. the effect of smoke particles’ concentration, the effect of smoke particles’ adsorption to the sample vessel, etc. And a few useful methods were taken to solve these problems. Two kinds of different cigarettes particles were measured in these experiments. The results of these two samples were different from each other and the results of each sample were fluctuated within a certain range, which denotes the application potential of DLS measurement of smoke particles in gas.

To solve the problem of local map building in the field of mobile robot navigation, we analyze advantages and drawbacks of several kinds of range finders, the model of laser range finder and its nonlinear problem which appears in local map building. A novel method is presented based on commonly model in autonomous mobile robot local map building. This method adopts weighted least means square to build local map and it can be easily implemented and has high precision. The emulation and experiment result shows that the method can effectively reduce the error of curve fitting in the field of robot navigation.

Based on the theory of radiometry, the measurement technique for linear response of ultraviolet ICCD (UV-ICCD) detector was studied and a comparative measurement method was presented. The surface of photocathode in the UV-ICCD was set in the same section with the end of fiber belonging to the reference detector. Firstly, the attenuation of irradiance was measured via reference detector. Then, UV-ICCD was set in center of the irradiance field. The outputs of UV-ICCD were recorded according to the corresponding irradiance at that time. At last, relationship curve between output gray value of UV-ICCD and input irradiance was given. Scientific-grade spectrometer with low non-linearity, which was less than 0.2%, was adopted in this project because the accuracy of the measure results depends on the accuracy of the spectrometer. Measurement instrument consists of standard deuterium lamp with high stabilization, optics-attenuator, integrating sphere, reference detector and computer. Auto control is realized via special software during the test. Experimental results show that non-linearity of UV-ICCD is less than 3%, and the uncertainty of comparison measurement method is less than 5%.

The error of measuring Brillouin frequency shift based on brim detecting technology has great influence on the measurement precision of sea surface temperature. According to the measurement principle of brim detecting technology and the relation between Brillouin shift and sea surface temperature, a theoretical model which monitors sea surface temperature by detecting Brillouin scattering frequency shift based on brim detecting technology is established. The effect of brim detecting technology on the Sea surface temperature measurement accuracy is analyzed. The results show that the maximal error of measuring Brillouin frequency shift is approximately 0.33 MHz while Brillouin frequency shift is distributed from 7.2～7.9 GHz, and the measurement precision of sea surface temperature is 0.010 5 ℃. This study has good instruction significance to the design of sea surface temperature monitoring system.

To promote the accuracy of Automatic Target Recognition (ATR) system, a new method was proposed for detecting airplane on high resolution remote sensing imageries, which had another procedure of clutter rejecting after target detecting in ATR. Firstly, we divided the whole image into small blocks based on the strategy of average area with superposition. Then the local recursive image segmentation algorithm was used to segment interesting targets on every block. With respect to the processes performed before segmentation and the fact that the segmentation of airplane could not be performed perfectly, an airplane’s axes detecting method was put forward by using added line and radon transformation. Through these processes, the recognition efficiency and the performance of airplane ATR were expected to be promoted much better. Finally, an experiment shows the potential ability of this new method especially for the under-segmented binary image of airplane.

A ship target detection scheme based on optical remote sensing images of contrast box filtering over the standard deviation feature planes is presented. Firstly, local intensity variance is selected as the detection feature to characterize ships of different intensities uniformly and eliminate the influence of varying sea average intensity at the same time. Secondly, on the two-dimensional detection feature plane, local detection thresholds are estimated self-adaptively by contrast box filtering and the spatial structure information of ship targets is combined to identify the locations of ship candidates in the intensity image. Thirdly, verification of each ship candidate is performed in combination with the prior knowledge ship feature model to remove the false alarms, and then the final results are outputted. Experiments conducted on the optical images demonstrated that the ship target detection probability can be as high as 99.5% and the false alarm rate is kept at 5%.

In order to improve the positioning accuracy of blind pixels in Infrared Ray Focus Plane Array (IRFPA), the principle of blind pixels appearance was analyzed in this paper. The environment temperature affected the detection capability of effective pixels, quantity and position of blind pixels. Therefore, the algorithm of blind pixels detection for IRFPA considering different environment temperature was proposed. The purpose of the algorithm was to position blind pixels and get quantity of blind pixels in different temperature ranges accurately. The results of experiment indicate that the algorithm can detect the changes of the positions of blind pixels with the environment temperature.

To simulate and calculate "the aero-optical transfer effects", an infrared imaging model was presented for applying the coordinate conversion method to trace any light in the space to figure out "the Optical Transfer Function (OTF)" and "the images". Based on the principles of geometry optical imaging and coordinate conversion, this method utilizes the density data based on the Computational Fluid Dynamics (CFD) grids of the hypersonic flow field to establish the model and emulate the blur effects of the field. And the results were contrasted and analyzed. The experiment results indicate that the model can reflect most of the known aero-optical blur phenomena in definite bound of precision. At last, we probe and discuss the precision and the significance of engineering application of the model.

According to the characteristic of infrared images, a new automatic fuzzy segmentation method was presented based on genetic algorithm for vehicle target image. Firstly, a region of interest was selected in order to reduce computation cost. Secondly, the region of interest was enhanced by fuzzy algorithm. Thirdly, 2D Maximum Between-cluster Variance algorithm was applied to segment the region of interest. At the same time, the genetic algorithm was combined with 2D MBV to make the calculation faster by its capacity of searching the best answer in a threshold bound. Then we detected fuzzy edge based on shortening width of fuzzy edge. At last, the final segmentation image could be obtained by OR and fill operations for the segmentation region by 2D MBV and the fuzzy edge. Experimental results show that only main body of the tank is segmented from the infrared image by one-dimensional and two-dimensional OTSU method. The new method can segment not only main body of the tank but also fuzzy gun from the infrared image.

The ZnO thin film and rare earth (La、Nd) doped ZnO thin films were deposited on Si(111) substrate by RF magnetron sputtering. The X-ray Diffraction (XRD) analysis revealed that ZnO thin film was highly c-axis orientation and RE-doped ZnO thin films were nano-multi-crystal departure from normal growth. The roughness surface figures of the films were observed by Atomic Force Microscopy (AFM). The room temperature Photoluminescence (PL) spectrum indicates that the thin films have strong purple peak at 395 nm and weak green peak at 495 nm. The PL spectrum peak intensity of Re-doped ZnO thin films is different. The results indicate that the peak of Nd-doped ZnO thin films is weakened and that of La-doped ZnO thin films is strengthened, and the causes of PL peak intensity changes are analyzed.

Because of the difficulty to fabricate the ultra-thin mirror individually, we proposed a base supporting method. Methods which were applied to control the distortion caused by adhesion, heat, and stress during ultra-thin mirror fabricating were researched, and means to estimate the effect was put forward. Technology of controlling mirror distortion was improved based on the experience on φ195 mm ultra-thin mirror manufacture. Then a φ340 mm ultra-thin mirror was fabricated by improved technics. When it was de-blocked from blocking body, the surface error of PV=5.74λ, RMS=1.02λ (λ=632.8 nm) caused by distortion is acceptable. Those experiments indicate that those technics can control distortion effectively and technical improvement is correct. Results of these experiments also indicate that the final error on ultra-thin mirror surface is astigmatism, which is not caused by gravity.

The effects of polarization degree, extinction ratio and bias angle error of polarizer on performance of comparative Optical Current Transformer (OCT), were analyzed in detail by employing polarization analysis tools. The theoretical analysis result shows the comparative OCT can eliminate the influence of nonideal polarization degree and extinction ratio because of the sensor’s dual-input and dual-output structure, while the bias angle error brings great damage to the performance of the comparative OCT. This disadvantage can be decreased by adjusting position between polarizer and analyzer exactly and increasing reference DC magnetic field or correcting software. The experiment of comparative OCT is carried out, and the test result shows that the change of error of the comparative OCT is only 0.7% in the range of 50℃ temperature, while the change of error of an ordinary OCT is more than 16% before comparative compensation. Moreover, the temperature characteristic of comparative OCT is much more excellent.

Digital video camera has the properties of large noise, uncertain location of main body and high requirement of real-time. The disadvantages of traditional auto-focusing algorithms were discussed in detail and a new auto-focusing algorithm was proposed. A novel focus measure filter called First Derivative of Gaussian (FDOG) was developed to estimate the degree of focusing correctly by the suppressing high frequency noise. To enable the auto-focusing system to focus on the exact main body, an automatic region-of-interest selection strategy based on the principle of nearest-subject-priority was adopted. Furthermore, combined with mountain-climbing search, binary search and quadratic interpolation, the proposed real-time search strategy greatly reduces the convergence time by about 40%, which enables the video camera to fast focus. The proposed algorithm is successfully implemented on a prototype digital video camera and experiment results are given to demonstrate the advantages of the proposed auto-focusing algorithm.