The TV guidance systems need to detect missile target signals from images in nature fields and complex backgrounds. Based on the different spectral characters between missile targets and backgrounds, the missile target’s spectroscopic detecting method and theory are proposed. And a spectroscopic detecting optical system with dual CCD sensor is designed. Its light-splitter let images including targets enter the primary CCD sensor and background images enter the secondary CCD sensor. Then a background subtraction circuit subtracts the secondary CCD sensor’s image from the primary CCD sensor’s image. The result is an image containing only the missile target information. Experiment result shows that the method is reliable in detecting missile targets from complex backgrounds.<收到修改稿日期2008-03-27

For the infrared real-time dim small target detection and tracking of the image sequences,a parallel processing system based on 4×ADSP TS201S DSP is designed. The loose coupling parallel processing system is constructed by the TS201S link ports. The system possesses Low Voltage Differential Signaling (LVDS) digital video interface, and supports cross interconnecting between DSPs and board level interconnecting. The computational performance of the system can be up to 19.2GMAC@16 bit/s in fixed-point operation and 14.4 GFLOPS in floating-point operation. The infrared dim small target detection and tracking methods have been implemented in the system. The experiment result shows that the parallel processing system has real-time ability, applicability and expansibility. And the real-time infrared dim small target detection and tracking of the large size and high frame frequency image sequences can be realized.

To solve the mismatching between candidate model and reference model caused by state changing of the tracked target, a novel model updated method based on object prior knowledge was proposed in Mean Shift framework. The algorithm was implemented by using an auxiliary model constructed on color statistical knowledge of the tracked object under different states. This auxiliary model instructed the update of different color characters. Therefore, a reference model with tracked object prior information was formed. In such way, the time selection problem of model update was also solved by the proposed method. Experiment results under complex scenes show that the novel algorithm overcomes the mismatching of models caused by head rotation effectively and is able to realize continuous head tracking. In addition, the adaptive ability of Mean Shift algorithm is greatly improved.

In a sensed image of long distance, the gray levels of target and background are hardly distinguishable, which results in a low-contrast image. Dim-target detection is always a difficult problem. The aim of this paper is to propose an anisotropic diffusion filtering algorithm based on partial differential equation to enhance the dim targets. The algorithm establishes a new filter model by improving the traditional P-M model based on the anisotropic diffusion theory. The proposed method adaptively performs the smoothing process in the faultless areas to make the background uniform, and performs the sharpening process in the variational areas to enhance dim targets. Simultaneously, we can select the smoothing and sharpening degree by adjusting the parameter K and w to satisfy different environments. Experimental results show the efficiency of the proposed diffusion scheme in dim-target enhancement with low-contrast.

In order to analyze the difference of optics sizing and aerodynamic sizing of aerosol particle, a method for revising the aerodynamic sizing using optics method is proposed. The method uses the measurement results by an Aethalometer (AE), an Optical Particle Counter (OPC) and an Integrating Nephelometer(IN)to revise the aerodynamic sizing of aerosol particle measured by 3321 in Xiamen. The comparisons with other instruments show that the method is satisfactory. Lastly, the reason for difference of optics sizing and aerodynamic sizing is analyzed.

A novel type broad beam cool cathode ion source was successfully developed for optical thin film growth technique by ion beam assisted deposition. The structure, discharge process and working properties were introduced in this paper. The ion energy and its distribution were investigated by a 5-grid tester. The results indicates that the lowest ion energy in beam increases with the increase of the extraction voltage, whereas decreases with the increase of the pressure. The peak position of the ion energy distribution function shifts to the low energy with the increase of the pressure, whereas to the high energy with the increase of the extraction voltage. The average ion energy is about 600～1 600 eV at the extraction voltage of 200～1 200 V, and the initial ion energy is 430～480 eV. The ion energy and energy distribution can be controlled effectively by regulating the working parameters of the ion source during the film deposition process.

In the 70° big visual field photoelectrictity measuring system,distortion produced by optical system makes the image of object in linear CCD deviate from the theoretical point and lead to the system measuring error. In order to improve measuring precision for the goal point in the measuring system,a set of device made up of eleven equidistance goal points for detecting optical distortion is designed on the basis of distortion theory. According to the change condition of systematic distortion,the compensation equation need fitting to join point location imagined by object point on linear CCD in real time measure. By the cubic equation fitting based on the data detected in measurement system as math model,the compensation program is compiled to carry out distortion correction in experiment program. The experiment result in actual detection shows that the device can reduce error obviously from -2.08 mm to 0.05 mm according to the compensation equation when the height is 200.115 mm,and raise the whole systematic detection precision from 1.039% to 0.025%.

In three-dimensional (3D) shape measurement based on phase measuring, the ultimate and important job is to map the phase to height and map the image coordinates to the global ones, which are usually done by 3D calibration. It’s a key process because of its relation to the accuracy of the measurement. Based on the dual-direction nonlinear phase-height mapping technique and the Pinhole camera model without lens distortion, and considered the characteristic of Fourier spectrum of a linear structured signal, a method which can complete the calibration of height (Z) and the lateral coordinates (XY) simultaneously by orientation filter is presented. The method has been used successfully to calibrate the system of rotating blade measurement, its standard deviation is less than 0.27 mm in the lateral coordinates (230 mm× 230 mm) and 0.022 mm in height (72 mm). 3D calibrations can be easily and perfectly done with the method.

To resolve the problem of detection of similar foreign materials in the complex background, a detection method based on transmission/reflection spectral imaging was proposed. According to the method, the different imaging model was constructed according to the different transmission/reflection characteristic discrimination between foreign materials and background. In different imaging model, the optimal illumination wavelength that distinguished the foreign materials from background was selected, and a transmission/reflection spectral imaging system was established. In actual application of detection of foreign materials in cotton, the image features of similar foreign materials from those of background were acquired by using optimal wavelength transmission/reflection spectral imaging system, and then the image targets were extracted by using adaptive thresholding image processing. The experiment result indicates that the detection ration is 91% under the condition of reflection illumination coefficient 0.74 and transmission illumination coefficient 0.83, and the method is a novel practical mean for detection of similar foreign materials currently.

Referring to limitation of the conventional mercury lamp spectral calibration in ultraviolet wave band, a spectral calibration device was built for ultraviolet spectroradiometer. The physical process and measurement chain of spectral calibration were studied and twelve ozone absorptive spectral lines were calibrated. The measurement uncertainty and combined standard uncertainty of influenced quantities of wavelength calibration were analyzed and the combined standard uncertainty was 0.026nm. Spectral bandwidths can reach 1.0nm according to the analysis of optical CAD and experiment results. The spectral calibration system does well in ultraviolet spectral calibration for spectroradiometer. The calibrated results satisfy the spectral precision for ozone retrieval.

The parameter error of inertial navigation system has important effect on the precision of navigation system. Conventional method of angular velocity calibration can’t identify the bias and the error calibration of the Fiber Optic Gyroscope (FOG) under the insufficient preheat condition is affected by temperature fluctuation. To solve the two problems, a new calibration algorithm for FOG Strapdown Inertial Navigation System (SINS) was brought forward, called Adaptive Recursive Least Square (ARLS). After analyzing the FOG Inertial Measurement Unit (IMU) errors and their compensation model, the method of adaptive forgetting factor was studied through large temperature tests and implementation procedure was given in detail. Finally, the validity of ARLS that can identify the error coefficient of FOG, and restrain the effect of temperature fluctuation in the condition of unstable device characteristics, is proved by algorithm emulation and angular velocity test.

We analyzed the stray light source of system and drew a conclusion that non-image beam has great effect on system qualities and system ghost images caused by parallel windows should not be ignored. In view of the facts that traditional graphical construction has less pertinence to stray light and tracing different incidence angle rays is fatiguesome and more likely to omit some rays, we proposed a reverse ray-trace method based on the ray reversibility. According to the method, a dot-matrix source was set at the image plane. Each dot emitted rays to fully cover the system aperture. A detector surface, large enough to receive all possible rays from the source, was set at object plane. Using the method, non-image stray light was easy to be recognized. Outside and inside lens hoods with special demands were designed,and its performance was simulated with software.

Based on the principle of testing aspherical surfaces with Computer-generated Holograms (CGH), a binary phase twin CGH is proposed and designed. It consists of two segments, namely main CGH and alignment CGH. Main CGH, fabricated on the inner part of a substrate, is used for testing an aspherical surface. The outer segment is alignment CGH, used for aligning the main CGH accurately by eliminating errors such as defocus, decentration and tilt on the premise that they are concentric with each other. A fabricating method which is able to manufacture twin CGHs with two different phase depth is introduced. Both holograms are fabricated on one mask simultaneously. After being exposed, the substrate is etched twice. Finally, a design example of twin CGH for testing a paraboloidal mirror (Φ140、F2) is presented. The validity of design method and the feasibility of fabrication process are demonstrated by comparing the test result with that of the autocollimation test.

Sample noise response method used to determine subband threshold factors of contourlet transform denoising is proposed. By the method, we can obtain the convergence factor of each subband according to every subband statistical character driven by a standard Gaussian white noise. The hard threshold of every directional subband of each scale is determined by modifying the 3σ (or 4σ) rule in terms of corresponding effect factor. The post-processing can be carried out by adaptive Wiener filtering followed subband hard threshold in contourlet domain. Experimental results show that, using the denoising method proposed, for images corrupted by Gaussian white noise, the denoising results including Peak Signal-noise Ratio (PSNR) and quality of visual effect are satisfying. To calculate the convergence factors using smaller sample image can accelerate the denoising speed and reduce the requirement of the memory for the program, and the denoising results can be kept satisfied.

A new class of noiseproof adaptive Bandelet bases (NA-Bandelet bases) is proposed. First, by employing soft-threshold function and the threshold obtained by SURE principle, the approximation of the original image in each dyadic square is computed, and then the geometrical direction in each dyadic square is found out according to the principle of the minimal approximation error. Secondly, the small dyadic square is combined based on the principle of Minimum Mean Square Error (MMSE), so the directions of NA-Bandelet bases are determined at last. Therefore, a novel denoising method based on the NA-Bandelet bases is presented, and it is applied to denoise and despeckle the optical image and SAR image respectively. Compared with the denoising methods based on Orthonormal wavelet bases and Bandelet bases, the experiment result shows that the algorithm proposed in the paper can reduce noise more effectively with improved edge-preserving ability.

A new reconstruction algorithm based on maximum entropy theory for two orthographic directions is proposed. The method can acquire a good reconstruction result, which just needs two orthographic projection data. The new algorithm can make experimental system simple and show great superiority to multi-peaks asymmetrical flow fields reconstruction because only two orthographic direction projection data is necessary. Single-peak random cosine function, two-peak random cosine function, three-peak random cosine function are used as tested fields in numerical simulation, respectively. We also consider the ability of the proposed algorithm to suppress the noise. Under the condition of Gauss noise, the numerical simulation results show that the root mean square error of single-peak random cosine function is improved by 62.8%, the root mean square error of two-peak random cosine function is improved by 63.3%, and the root mean square error of three-peak random cosine function is improved by 58.6%.

In view of lidar return signal easily polluted by various noises, a new method of Empirical Mode Decomposition (EMD) that can analyze the nonlinear and non-stationary signal is introduced to decompose lidar signal. The method based on signal intrinsic time scales is used to extract the energy and frequency of signal. With the method, the Mie lidar signal polluted by noises is decomposed into a number of Intrinsic Mode Function (IMF). In the reconstruction course of IMFs, the IMFs including high-frequency noise are excluded to reduce the noise. The experiment results demonstrate that the method has powerful and adaptive de-noising ability. To denoise Mie lidar signal by using the method has been attempted. It is believed that the method can provide a new effective way of noise reduction for weak lidar signal.

A novel image texture exaction algorithm using Independent Component Analysis (ICA) filters for image classification is proposed. Firstly, image patches are selected randomly from images in training set as observation signal. A group of filters (ICA filters) is extracted from the sample texture images using ICA method. And then, ICA filters are evaluated and selected according to the response of the input sample images for the purpose of reducing feature dimension. Finally, response results of test images are gotten by using the filters to the images. Global and local features are extracted from the histogram of the maximum response results. Experiments are completed on the images with 108 texture classes from the Brodatz album. Experimental results show that the proposed texture feature has better classification rate than that of MPEG-7 texture descriptors.

A new method of sub-pixel localization based on SUSAN and Hough transform was presented. The size of template selection was proposed and the straight edge response was defined in SUSAN algorithm. Meanwhile, weighted Hough transform was conducted. First, the straight edge response was used to extract edge. Then, the Hough transform was computed only when edge points of line had the value of the response and the value was added in an accumulator as a weight. The local areas of parameter space were divided for the second Hough transform and the edge detection was attained by point fitting as sub-pixel localization position. Experimental results show that the sub-pixel localization precision of the method can reach 0.3 pixels. It has greatly decreased calculation compared with traditional edge detection and Hough transform. The method is simple and has better properties of anti-noise capability and robustness. Meanwhile, it provides a new way for sub-pixel localization of analytic curve.

Concepts of boundary flow and contour flow, and a method of edge feature extraction, are used for object detection and recognition in SAR (Synthetic Aperture Radar) image. The method first used medial filtering, automatic thresholding, mathematical morphology binary filtering, and cross templet edge detection to extract pixel-based boundaries. Then, contour following was carried out and boundary flow entropy and contour flow entropy were computed to obtain the boundary flow features and contour flow features of target edges. Using the method, experiments of edge feature extraction of test images and SAR images had been done. The experimental results show that this method can describe the feature of objects well and also has the advantages of simple programming, faster computation, etc.

Aiming to the question that geometric attacks desynchronize watermark detection to fail many watermark schemes, we present a new geometrically robust image watermarking scheme based on Tchebichef moments (GWTM). Geometric distortions are first resynchronized by image normalization, and watermark insertion and extraction are implemented in normalized domain. To select a moment set fits watermark embedding using the property of Tchebichef moments. According to the magnitude changes of the above moment set after being modified, an optimal moment set is determined by test, and the moments within it are modified according to the random watermark bits. The modified optimal moment set is reconstructed to obtain the watermarked image. In detection, the watermark bits are extracted by comparing with the Tchebichef moments of original image with the optimal moment set as a key. Experimental results show that GWTM can efficiently resist rotation attacks of random angles, scaling attacks and common signal processing attacks.

In view of the problems of feature extraction in face recognition, a new face image feature extraction and recognition method—Gabor Discriminant Locality Preserving Projections (GDLPP) is proposed in this paper. GDLPP first gets the high-order statistical information by calculating the Gabor wavelet representation of face images. Based on LPP, GDLPP takes into account the inter-class information, changes the objective function, and extracts the discriminant feature of face for recognition. The proposed method was tested and evaluated in the USPS database, Yale face database and AR face database. Nearest neighborhood algorithm was used to construct classifiers. The experimental results show that GDLPP has good performance even if pose, illumination, face expression and train sample number change.

A fast and robust front-view object location algorithm based on point pattern matching is proposed. It transforms the problem of front-view object location to the problem of matching between a 2-d point set and a 3-d point set. The matching matrix and the transformation parameter are solved through minimizing the object function restricted bi-directionally by matching one to one in the two point sets. The fuzzy degree of matching matrix is controlled by the temperature of deterministic annealing algorithm, which improves the robustness of the algorithm and reduces the possibility of getting trapped in local minima. The effectiveness and robustness of algorithm are indicated through experiments.

A new auto-focusing algorithm based on Depth from Focus (DFF) was presented according to the properties of the automatic iris imaging system. The algorithm used the ULFM value of the image as the sharpness evaluation function on the basis of the fact that the automatic iris cameras have small depth of field, and employed a two-step auto-focusing strategy by combining coarse focusing with fine focusing. Experiments of automatic iris imaging system prove that the auto-focusing algorithm proposed is of high precision and strong robustness. The searching time meets the needs of real-time auto-focusing. The system can accurately acquire clear iris images for recognition with a work scope of 0.5 meter to 2 meters and the rate of successful focusing is higher than 95%.

To reduce the influence of eyelash noise and improve discriminability in iris recognition, a detection of eyelash occlusions method based on the canny method was presented for the iris recognition. The parameters in canny method were researched. One sector in the outer circularity of iris is used to represent the region of eyelash occlusions. Excluding the pupil and the sector, the rest of the outer circularity is the effective region of iris. Our experimental results demonstrate that the sector contains all eyelash pixels in the outer circularity. The algorithm can detect the eyelash effectively and can improve the classification and discriminability of iris patterns and the rate of iris recognition.

The structural design of the R-C telescope system is researched. We utilized finite element analysis in the whole structure of R-C telescope system. Finite element analysis results of modal analysis indicate that the first natural frequency of the telescope is 183.39 Hz, which is higher than 100Hz requested by the system. Finite element analysis results of the telescope with temperature load indicate that the coaxial degree of primary-mirror and sub-mirror is 1.15 μm, which is less than 0.01 mm requested by the system. The theoretical results are coincident with experimental ones, which validates the rationality of the R-C telescope’s structure.

The lowpass filter in the CCD readout circuit inevitably results in information distortion of pixel crosstalk while improving the system signal-to-noise. To solve the information distortion caused by the pixel crosstalk in the CCD readout circuit, the relation between the information distortion and the cut-off frequency of the first-order lowpass filter was deduced. On the basis of information distortion theory analysis, a new method of digital compensation first-order lowpass filter was proposed to resolve high order filter’s realization inconvenience. The proposed method adopted first-order lowpass filter to design the CCD readout circuit. The designed first-order filter may greatly reduce the cut-off frequency and at the same time satisfy the system information distortion degree. So the designed first-order filter may be substitute for high order filter in the application system. Finally, the validity and feasibility of the method are verified by the elaborate experiment.

The absolute position and intensity are fundamental parameters for the synchrotron radiation beamline. An double-wire X-ray Beam Position Monitor (XBPM) was manufactured to solve the problems in BSRF-3B3 diagnosis. We stated the significance of developing a double-wire XBPM and introduced the optical principle, physics design, mechanical design, data collection and operating effect in detail. At BSRF-3B3 beamline, an XBPM system was developed for the absolute position and intensity detection. The information obtained from the XBPM system is guidance for adjusting the B orbit parameters, and measuring vertical position and Full Width at Half Maximum (FWHM). The XBPM system plays an important role in the performance optimization of BSRF-3B3 beamline.