An optoelectronic measurement method for real-time monitoring deformation of the large-scale platform is presented. This method place some linear array CCD as a high benchmark test at various location for measurement, and makes use of the laser as line source so that the linear array CCD of different positions receives the same signal of beam of light at the same time. Then, linear coordinate transformation is used to transfer location information of height which linear array CCD receives into elevation output whose baseline is coincident with other coordinate. Compared with each high value data in different load conditions or at different moments at last, real-time relative deformation of points is obtained in the platform. The method brings about good results in practical application.

An optical mapping method for acquiring CAD model of turbine blade and its tiny holes is proposed. Firstly, making use of Digital Structure Light (DSL) 3D scanner of small field of view, the multiple profile segments of the blade are measured and integrated into the whole profile through alignment. Secondly, some high accurate cylinder gauges of proper diameter are inserted into tiny holes and point cloud of gauges are measured and integrated into the blade profile. Finally, NURBS surface is fitted by point cloud of the blade and the cylinder surfaces are fitted by point cloud of gauges. The positions of tiny holes are intersections of cylinder axes and NURBS surface, and the normal orientations are cylinder axes. The method measures gauges rather than directly tiny holes through transferring measurement object. The high precision and reliability of the reconstruction model are reached by increasing sample density of measuring point cloud. The mapping result shows, the fitting error of NURBS surface and cylinder surface are both 0.104 mm.

A new method for measuring the vibration frequency of a vibrating object was put forward. The experiment system was put up and a lot of experiments were made. The system can give out the vibration model of a vibrating object by receiving the scattered light of the vibrating object and then making an auto-correlation of the scattered light. Moreover, the system gives out the vibration frequency spectrum by doing a Fast Fourier Transform (FFT) to the auto-correlation data. Adopting optic-fiber to transmit the scattered light, adopting a photomultiplier as a photon detector and adopting a data processing technique combined FFT with auto-correlation analysis, the system can work well without any special request both to the light source and the measured object. Furthermore, the system has a wide range of application, being simple to operate, and is easy to be used in non-contacting, long-distance and network way. The simulated experiments indicate that the system works well in a wide frequency range of 1 Hz to 105 Hz.

In order to improve the centering precision of non-diffraction laser beam, a sub-pixel edge detection method of bright ring is proposed. The pixel roof edge points of a bright ring were detected based on maximum intensity in a local region firstly. These pixel edge points were then processed by linear interpolation to get sub-pixel edge points. Finally, the center of the laser image was calculated by using the least square method of all these sub-pixel edge points. For 50 samples captured (with a 640 pixels×480 pixels camera and 0.057 2 mm/pixel as resolution of the target) at each position at different distance of 40 m, 50 m and 60 m, the standard deviation for center fitting of the method with sub-pixel edge points was 0.021 4 mm, 0.037 2 mm and 0.042 3 mm for x, and 0.022 8 mm, 0.035 1 mm and 0.043 7 mm for y, respectively, much better than that with pixel edge points. Experiment results show that the method is feasible.

According to characteristics and demand of fine tracking spot detection in space laser communication system, a kind of real-time spot detection instrument was brought forward by adjusting gain automatically , which can overcome saturation photic intensity effect. Function of calculation center of spot, smoothing filter et al. was implemented by using DSP for spot change. By experiments, miss distance can be exported in real-time, and gain can be adjusted automatically with the change of input powers, which can overcome the failure brought about by saturation photic intensity. Shift characteristic curve of four-quadrant detector is measured, which has been proved that 10.1 μrad of angle resolution has been obtained. Furthermore, the ability of subdivision can be double by digital filtering. The parameters of filter order, orthogonal compensation angle etc. can be set from external configuration.

The method of dim target detection based on spatio-morphological and wavelet transform clutter suppression is proposed. The image sequences are processed with spatial tophat filtering. The results are transformed into wavelet domain, and filtered by smooth filter in every wavelet belt, and the difference process is operated between the pre- and after filtering coefficients. The inverse wavelet transform is carried out to produce the residue image sequences with dim targets. The tophat filtered image is used to estimate the possible target support area. Under the restriction of the possible target area, the tempo-spatial data fusion is performed to detect the trajectories of dim targets. Simulation results show that the obtained residual images have very good white Gaussian normality, and the average gain of the target’s neighborhood Signal-to-clutter-noise Ratio (SCNR) can be improved obviously compared with traditional image smooth filtering and frequency filtering algorithms. The trajectories of dim targets can be detected steadily with 5 image frames.

An improved Multiple Model Particle Filter based on Track-before-detect (MMPF-TBD) algorithm for maneuvering target detection and tracking in low signal-to-noise environment is proposed. The algorithm uses sliding window to determine whether the particles are affected by the estimation of the target. When the value exceeds threshold, it adds new particles in accordance with the state estimation of the previous moment. Then it uses the expanded particles to detect and estimate the target. Compared with the existing methods, the simulation results show that the proposed algorithm can effectively solve the particle degeneration problem and improve the probability of maneuvering target detection and tracking accuracy in real time.

Position accuracy of star sensor is restricted by color temperature of star. In order to analyze the effect of color temperature in star sensor, the model of stellar spectrum was presented in this paper. And the position errors of star, which were caused by different stellar spectra, were calculated in an optical system of star sensor. The results show that the star position errors are not remarkable near the optical center. And the effect of color temperature could be minimized by reducing the working wavelength range of optical system, but the spot intensity would be lower at the same time. For example, if the working wavelength range is reduced from 300 ~1 100 nm to 400 ~800 nm, the single star position errors will be 0.042 2″、1.965 2″ and 3.389 1″ when the field view are (0°,0°), (0°,2°) and (0°,4°) correspondingly, which are about 54％、65％and 70％of the original errors. And the spot intensities are about 58％, 62％, 63％ and 51％ of the original when color temperatures of star are 9 600 K, 7 600 K, 5 600 K and 3 600 K correspondingly.

Photon-counting detection is a key technique for detection of weak intensity light and high-speed imaging study. A photon-counting detection and imaging experimental setup was designed and built. Its structure and working principle were introduced in detail. The experimental hardware was composed of Avalanche Photodiode (APD) arrays, suitable driving and quenching circuits and NI-PCI high speed data acquisition card. System software and data acquisition display interface were programmed with Labview and VC++ combination. Weak photon irradiated on the avalanche photodiode which works on Geiger mode was collected by high speed data collection card continuously and processed with personal computer software. Finally, the photo-counting pulse image was acquired by this system. The experimental setup provides a powerful tool for very faint luminescence detecting and high-speed imaging study.

Through the interferencing light intensity under a sine-wave modulation signal analysis, the output signal of fiber-optic gyroscope is demodulated. And then the change of the output signal of gyro is gotten. Through the analysis of the change we know that the effect of residuary intensity modulation on the performance of Interferometric Fiber Optic Gyroscope (IFOG) is related with the precision and accuracy requirements of IFOG. The higher precision, the more obvious effect. Secondly, we analyze the digital ladder height change caused by the demodulation output voltage change in the condition of the effect of residuary intensity modulation on the digital closed-loop fiber-optic gyro. Moreover, through the residuary intensity modulation coefficient and digital ladder height formula, explain the influence of residuary intensity modulation on the fiber-optic gyro zero bias and scale factor. Finally, we simulate the relationship of the residuary intensity modulation coefficient and the scale factor stability of fiber-optic gyro by using the Matlab simulation tool. The simulation results further explain the effect of residuary intensity modulation on digital closed-loop fiber-optic gyro scale factor stability.

The self-focusing properties, design, fabrication and diffraction properties of Varied Line-Space (VLS) grating were investigated in detail. The VLS grating was automatically optimized and generated by the macro program. The 300 l/mm transmission VLS grating operating at 355 nm was fabricated by e-beam lithography. The diffraction properties of VLS grating were characterized with focusing-spot testing system. The diffraction peaks of fabricated VLS grating were measured with a frequency tripled Nd:YAG laser. Moreover, the experimental results were compared with those using He-Ne laser. It has been indicated that the VLS grating of active design has promising ability to significantly improve the diffraction intensity and spectrum resolution.

Heterodyne efficiency is reduced by aberration of receiving antenna for space coherent optical communication. On the assumption that local oscillator wave is Gaussian distribution and signal wave is Airy distribution, a simple analytic expression of heterodyne efficiency is given for an aberration-free system. When the ratio of signal Airy disk radius to local oscillator Gaussian 1/e field radius at focal plane is 1.71, maximum heterodyne efficiency of 81.45% can be obtained. The effect of signal wave receiving antenna’s tilt, defocus, spherical, coma and astigmatism aberration on heterodyne efficiency is analyzed by assuming that local oscillator wave is perfect Gaussian distribution. One-dimensional integral expression of heterodyne efficiency including aberration effects is given. The result shows that one wavelength spherical aberration could induce 0.9 dB heterodyne efficiency loss even by using defocus adjustment. So spherical aberration should be carefully considered in designing an on-axis receiving antenna.

To meet the requirements of a Computed Radiography (CR) and to implement its functions, a new high-performance laser scanning system was designed. A scanning arm was designed which was composed with pentagonal prism and scanning object lens. Scanning arm is used to replace the Fθ lens. To cooperate with scanning arm, a circular imaging plate transport mechanism was designed. Scanning arm can receive stimulation luminescence as a receiving device, too. The system designed by this method has simple structure and good performance. Besides, the control scheme was introduced and timing sequence diagram of control system was given. The Influence of pentagonal prism’s displacement errors on system’s resolution was calculated. The calculation results show that this system is easily adaptive to the error. Experiments were carried out to verify the performance of laser scanner. Moreover, the results indicate that the images have good visual effect and can meet requirements of industrial inspection.

The backlash of gear influences the pointing accuracy and tracking accuracy. Backlash could be eliminated by applying bias torque in multi-motor driving system. Dual bias driving system was taken as research objective. The mathematics model of dynamics was established, and many nonlinear factors which might affect the dynamic behavior were considered including gear backlash, stiffness of the gear train, random wind disturbing and cogging torque of motors. Minimum of bias torque was taken as the objective function. The driving gear and driven gear didn’t separate during work, which was taken as the bound for objective function. Moreover, the optimum bias torque under every driving torque at full range of speed was calculated. The changing rules of the optimum bias torque charged by each gear train parameter including inertia, stiffness and damping were analyzed and achieved.

According to the design targets of a mirror which is made with the material SiC in a certain space optical system, analyze and contrast the calorific capability (which is figured by the deformation caused by the different temperature) and the mechanical capability (which is figured by the deformation caused by the gravity) of many different light-weighted structures (such as triangle, square, hexagon and various sectors). Then, decide to use one of the sector light-weighted structures after contrasting and balancing the light-weighted percent, rigidity ratio (which is the ratio of the elastic model-E and the density of the material) and the calorific stability. With the sector light-weighted structure, the light-weighted ratio reaches 75 percent, and the deformation brought by the gravity and the calorific deformation are both enough small for use. The RMS of aberration-deformation of the mirror surface with surrounding support is 0.010 1 μm which is about λ/60 (λ=0.632 8 μm). The aberration-deformation does not contain the rigid displacement of the mirror surface which is removed by using the Zernike polynomial to fit the deformation of the mirror surface calculated by the MSC software.

An oral cavity endoscope was designed to meet the practical needs of dentistry. The whole system was composed of imaging system, illuminating system, image display and processing system. The system has the function of micro-amplification, and the brightness and color temperature is adjustable. In this paper, optics software was used to design and optimize the optics imaging system, and gave out the aberration curve. The illumination system was constructed. In the ending of the papers, the oral cavity endoscope was used to test two teeth model. The system resolution is higher than 100 lp/mm and the image is clear. The details of tooth surface can be observed. The experiment results show that the system could meet the requirements of dental medical application.

For registering multi-sensor remote image automatically, a new kind of virtual point based on line feature is proposed as a basis of point features and control points. Taken optical images and SAR images of airport as example, a multi-sensor remote image registration algorithm based on virtual points is proposed. The common features shown in two kinds of remote images, i.e. line features are extracted in different ways as the characteristic of each image firstly. Then virtual points are constructed on the basis of line features in each image, and used as the control point candidates in the process of image registration. The automatic coarse-to-fine registration between the two kinds of remote images is realized by two steps: the coarse registration based on feature consistency and the fine registration based on virtual points. Experimental results show that the proposed algorithm is robust and accurate, and realize the coarse-to-fine registration between optical and SAR images automatically.

To verify the validity of image compression algorithm, 122.0-B-0, we implemented it with software after a detailed research, and then compared it with two other classical image compression algorithms: JPEG2000 and SPIHT. The experimental results show that CCSDS algorithm has an equivalent performance to JPEG2000 and SPIHT when compressing in a low bit rate, but a little bit lower than JPEG2000 and SPIHT when compressing in a higher bit rate. The encoding speed is 2 times faster than that of JPEG2000, 1.5 times faster than that of SPIHT when compressing in a given bit rate, and the speed is directly proportional to bit rate. CCSDS algorithm uses a simple encoding method withour feedback operation, which can be used in image compression system with alterable memory, and also can effectively prevent error propagation. Therefore, it will be a great value for application on spacecraft.

Aiming at drawback of high computation complexity of multi-scale Retinex, an algorithm for contrast enhancement of image which simulates visual characteristics of human eyes is proposed. According to Retinex model, the image is separated into illumination and reflection component. And two nonlinear transformation functions are introduced to modify illumination and reflection components of the image respectively. And the global contrast enhancement function that adopts logarithm transform could stretch illumination component, so whole visual appearance is improved. Then reflection component is adjusted by nonlinear function that involves discriminating ability of image gray level for human eye, consequently local contrast of image is enhanced. Experiments show that our method does not cause color distortion for RGB and some chroma-luminant color spaces, visual appearance is well, and operation cost is lower than that of Multi-Scale Retinex (MSR).

To solve the ill-posed problem of the super-resolution image regularization reconstruction, an energy function based on p(x) harmonic mapping regularization and super-resolution image observed model was drawn. The super-resolution image was obtained by a dynamic partial differential equation. The algorithm could adopt the different norm adaptively during regularization reconstruction, which used near 2-norm in smooth region and near 1-norm in the image edge region. The experiments show the algorithm not only reconstructs the super-resolution image efficiently, but also improves the “blocky” effect while preserving the edges.

SUSAN corner detection is a classical algorithm, but has not the property of scale change. Using multi-scale characteristics of wavelet Transform, positioning in small-scale is accurate and anti-noise in large-scale is strong. The multi-resolution idea is introduced into the SUSAN algorithm in this paper. Then, a novel SUSAN multi-scale corner detection algorithm based on wavelet transform is presented, which might detect the corners in different scales and overcome the drawback that the single-scale SUSAN detector usually leads to detecting false corners due to noise. For comprehensive utilization of various scales corner information, corner screening programs from coarseness to fine is presented. Compared with SUSAN algorithm, the presented algorithm is more efficient in detecting the corners with accurate location and is robust to noise.

The two thresholds of classical Canny operator need to be set manually, which limits the application of this algorithm. Therefore, many researches about how to choose threshold adaptively are done to solve this problem. Based on the gradient histogram, a method of threshold-adaptable edge detection is proposed. This method is on the basis of gradient histogram difference diagram with adaptive image classification techniques. It not only automatically sets the two thresholds, but also avoids disconnected or false edges in detection. Experiments prove that the method is threshold-adaptive and advantageous for edge detection in color image whose pixels of larger gradient amplitude are mainly located in the edge between the target and background.

An effective de-noising method in edge protecting was proposed to overcome the limitation of the current image de-nosing methods, which combines edge detection with neighborhood weighed window. This algorithm decomposes noisy images using stationary wavelet transform to keep phase invariance. Then it detects the edges of low frequency sub-band and high frequency ones, and gets the approximate information of the edge of original image by fusion the results of edge detection. Based on intra-scale dependency and directivity of wavelet coefficients,the method filters different sub-bands coefficients using corresponding forms filtering templates with weighed window, except the information of edge. A better restoration of image is demonstrated in the results of experiments, with detail of images kept as well as image noises decreasing.

A new image registration algorithm is presented based on contour-feature, which involves extracting the contour of the image, calculating the normal angle of each contour point and proceeding normal angle histogram statistics,then estimating quickly the rotation angle between two images by circular correlation of normal angle histograms. The estimation of rotation angle can simplify the estimation of other parameters and realize the quick registration. The proposed algorithm is invariant to image rotation, translation and scale, and robust to contour deletion and noise, which is extensive to be applicable to open contours and closed contours. Moreover, the proposed algorithm has been successfully applied in Printed Circuit Board (PCB) defect inspection to realize quick and exact registration of reference image contour and the testing image contour.

Factorization algorithms are widely used in projective reconstruction, which has an inherent advantage of being able to handle any number of images simultaneously without special treatment for any subgroup of views. However, these algorithms assume that all object points are visible on all images. In order to overcome the limitation, we present an algorithm to estimate projective shape, projection matrices, projective depths and missing data iteratively. Estimation problems of projective shape and projection matrices are solved in terms of singular value decomposition. According to the fact that the sum of linear subspaces, each of which is spanned by the rows of all points in each image, is equal to the linear subspace spanned by the rows of space points, projective depths are estimated. Experimental results with both synthetic data and real images show that the proposed method has small reprojection errors, good convergence property and practicality.

Referring to the limitation of driver face detection algorithm based on single feature in detection precision and reliability, a novel fusion algorithm of driver face detection is proposed. Firstly, an improved face detection algorithm based on Haar-like feature is used to detect the possibly existing face region in the whole image. Then, the detected region is extended adaptively and a face detection algorithm based on skin color feature in YCbCr space is used to detect the face again in the extended area. Finally, double matching verification is made by the defined area coincidence degree and geometric prior knowledge of human face and fusion detection of driver face region is achieved by establishing relevant location rules. Experiment results in various complicated road conditions show the effectiveness of the proposed algorithm.

In order to solve non-consistence color appearance between different viewpoint images in various imaging system, a color correction method of multi-view images based on multi-modes fusion was proposed. Four general linear and non-linear color correction modes were defined. Firstly, disparity estimation was performed between reference and source images. Then, by using the matching blocks as color matches, correction parameters for four modes could be determined. Finally, modes were fused by least-square error method to get the final corrected image. Experimental results show that the proposed method can achieve minimum root mean squared error between reference image and corrected image, and obtain good correction results.

A novel selective ensemble learning algorithm is proposed based on Moore-Penrose pseudo-inverse. After individual learning machines had been trained independently, a set of initial weights were assigned to the individuals. Then an method based on Moore-Penrose pseudo-inverse was applied to optimize these weights for a minimum generalization error, and only those with high weights would be chosen for the final ensemble. Experiments show that this method is efficient in the computation complexity and easy to be implemented. Moreover, the method can achieve higher generalization ability with a much smaller ensemble size than other popular ensemble approaches.

A novel license plate localization algorithm is presented based on the saliency map. The feature maps with respect to intensity, color and orientation are firstly extracted from the input color RGB image by using the bottom-up visual attention model. Moreover, these feature maps are combined into a saliency map. After the binarization of saliency map, several plate candidates are obtained. Then, vertical edges, the aspect ratio, and hopping frequency of the brightness of the license plate region are used to eliminate the undesired candidates and determine the real license plate region. The comparison experimentations show that the proposed method can reduce the pseudo-plate number due to the background environment and have strong environmental adaptability, anti-noise ability as well as high localization rate.