Pixel-based MOG background modeling is not a good method to deal with moving target detection in dynamic background. It is hard to model the dynamic background in this method due to the complex movement of background pixels. Thus, huge sums of false positives would occur. This paper aims to suppress the false positives in space domain and time domain. In space domain, class-prior probability and class-conditional-probability of a pixel were calculated from MRF model and Mixture of Gaussians (MOG) model. By combining the class-prior probability and class-conditional-probability, foreground segmentation was completed to suppress a majority of false positives of small size. In time domain, three target characteristics were observed: motion constancy, motion saliency and area stabilization.The result of experiments show that the false positives on dynamic background could be suppressed to a great extent,meanwhile, good detecting accuracy is ensured.

According to the characteristic of general expanding target, linked with the fact of some especial rapid expanding target (such as these targets that re-entry atmosphere rapidly or super rapidly), these targets’ disaggregation and characteristics were analyzed in the process of its violent changes, and it is pointed out that general image tracking algorithms based on contrast-based tracking algorithm were difficult to be applied in this case. Based on comprehensive features of these algorithms,an weighted-edge tracking algorithm is given. Combined with FDEKF filter, it realized stably and automaticly tracking the disaggregation of these targets which re-entry atmosphere rapidly or super rapidly. Through simulation testing with a sequence of images of these targets back into atmosphere, and compared with other contrast-based tracking algorithm, the results show that the algorithm has better stability and higher precision in tracking such targets with rapid running movements and radical changes, and it is easy to be implemented.

For tracking and measuring maneuvering target for video image, Extended Kalman filter (EKF) based on local linearization of KF is easy to be realized , but only has performance in Gaussian and mild nonlinear environment. An algorithm based on Unscented Kalman Filter (UKF) and Interaction Multiple Model (IMM) is proposed for maneuvering target tracking in complex nonlinearity environment or tracking small object in video image. The simulation results show that the tracking performance of UKF is much better than EKF in complex nonlinearity environment, and the third order tracking precision can be achieved.

To satisfy the requirement of detecting small targets in infrared image, an infrared targets detecting arithmetic is put forward based on Scale Invariant Feature Transform（SIFT）and weighted entropy. The method considered the imaging character of small targets. SIFT feature is used to extract special points and the possible location of target is obtained through the match between frames. Weighted information entropy is applied to find feature of gray scale information and entropy. Then potential targets are judged again by the features which have been calculated for the need of real-time dim small targets detection in heavy cloud clutter and complex infrared backgrounds. Experiment indicates that the method has an advanced practicability, and the performance of robustness is good in the sky background.

An algorithm for detecting the small infrared target was proposed. Analyzing the morphologic Top-hat operator,it is found that the small target can not be accurately detected only by relying on the Top-hat operator. Based on the Top-hat operator, a method of blocked adaptive threshold and target detection in small range are used to detect the small infrared target. And the Matlab simulation shows that the method can raise the success rate of detecting the small infrared target in a certain extent. By analyzing the result of two scenes, the method not only adapts in a single environment, but also has certain robustness in practical complex scene.

In order to expend the application of the laser diode linear array in the far field, the collimation system is required. A collimation system is designed by vertical double half micro-cylindrical lens combination array. Ray-tracing method is used to analyze the transmission equation of the laser beam and the divergence angle changing with some parameters of the lens is analyzed to get the optimization design. Then, the shaded model picture and the detector viewer are simulated in non-sequential tracing by Zemax-EE, and the laser spot is gotten which is close to rectangle. The incoherent irradiance concentrates in the center-weighted area, and the divergence angle of the beam after the collimation system is about 3 mrad. The collimation system can collimate the diverse angle of two directions simultaneously, and the fabrication and installation is easy.

Based on Night Vision and Electronic Sensors Directorate (NVESD) Target Acquisition Performance (TAP)models, the clutter impact on human-in-loop TAP is analyzed. Aimed at characteristics of laser jamming image, a modified TAP model based on clutter metric is applied to analyze laser jamming image. Then, an evaluation method for laser jamming effect based on Probability of Edge (POE) metric is proposed. The method combines with human target acquisition task and laser jamming effect evaluation is researched from the aspect of laser jamming influence on human search and detection probability. Experimental results of laser jamming to CCD show that the method can reflect the influence of laser jamming image on search and detection probability and can reflect laser jamming effect evaluation effectively. Hence, the validity of this method is verified.

For enhancing stability of the Nd:GdVO4 linear cavity blue laser, the thermal lens effect of Nd:GdVO4(gadolinium vanadate) crystal was analyzed with diode end-pumped. It focused on the relationship between the output power and thermal lens focal length of 447 nm blue laser. By improving the method of dynamic stable resonators and using MATLAB and the Matrix Laser, the parameters were optimized such as length of the resonant cavity, the radius of curvature of the input/output mirror and location of the various components in the cavity. The result of the imitation and experiment shows that the cavity has a good dynamic thermal stability.

This article describes the attenuation unit of laser energy of wide range, high precision and stepless in the semi-active laser guided weapon simulation system, which consists of two prisms (different sizes), precision displacement platform，controller and grating. The optical energy regulation principle is based on Lambert law. Only by changing the material thickness which the beam passes through，the light intensity could be changed. It achieves a wide range, high accuracy, high dynamic, continuous and stepless adjustment of laser energy. Compared with the traditional laser energy regulation system, it solves these problems such as the energy dynamics of small range, unsatisfied linearity, needing to multi-stage cascade and so on, and also has a simple structure, high stability, and low processing costs. We take 1064nm laser as the experiment, gives the design scheme and its implementation process. The method plays a role in promoting this technology.

In order to fulfill the observations of near Infrared (IR) CO2 absorption bands in reflected sunlight, a kind of high spectral resolution near infrared CO2absorption bands imaging spectrometer was designed based on the principle of imaging spectrometer and the theory of optical design. A simple planar reflection grating spectrometer approach consisting of an entrance slit, a two-element collimator, a planar reflection grating and a two-element camera lens was presented. Common glasses were chosen and aspheric elements with simple conics were allowed with a view to minimizing the number of optical surfaces. The optical system of the spectrometer was optimized by using ZEMAXoptical design software. The analyzed results show that the spectrometer’s spectral range is 1591-1621nm, resolution is <0.08nm, F-number is 1.8, and these all satisfy the requirements.

The spectra picture of infrared spectra in entire spectrum band can be obtained with the Fourier transform infrared spectroscopy, but sometimes only some one of the spectrums is concerned. Therefore, we need to do some spectrum thinning. The method of Chirp~~z transform was used in infrared spectrum thinning, which makes the local ranges of the spectra thinned, and some of these local ranges of infrared spectra pictures were enlarged effectively. This method could realize thinning of the infrared spectra picture in any kinds of band without changing amount of the sampling points. Because the sampling resolution of these local ranges was improved, the higher precision of infrared spectra pictures could be acquired. Both the theoretical analysis and experimental results show that the method of Chirp~~z transform has characteristics of high accuracy, high speed and strong flexibility, so it can be widely applied in lots of engineering projects.

In order to improve the transmission performance of the conventional Mach-Zehnder interferometer, a novel interleaver based on nonsymmetrical Mach-Zehnder interferometer with adding optical fiber ring resonator to interference arm was presented. The output expression was established and described by using the principle of fiber transmission and the matrix transfer function. Through numerical simulation, the influence of the coupling ratios of the optical fiber ring resonator and the length difference of interference arms on the different bandwidth interleaver was analyzed. The analytical results indicate when the length of interference arms is equal, the port output curves are quite ideal by taking some certain values for the coupling ratios of the optical fiber ring resonator. Between 3 dB points, odd channels have more than 15 GHz passband for 10 Gb/s and the other even channels have more than 60 GHz passband for 40 Gb/s. This kind of interleaver has more advantages as multiplexer/demultiplexer during upgrade from 10 Gb/s to 10 Gb/s plus 40 Gb/s.

Aimed at the abundant information about inner star dynamics on eclipse picture, a method for dealing with the picture was presented by contour line. According to blackbody or grey body radiation theory, the temperature is nearly equal if the brightness was the same. The color picture turned into binary processing with getting the threshold brightness value. Then it got a boundary line----the contour line on maps of appointed color rank by differential calculus twice. Solar eclipse picture is marked by selecting special contour line on maps. The sun surface spray, intensely hot atmosphere current vertical movement or advection are obtained from the solar corona, and the shape, structure and developing trends of the particle beams and solar prominence, and even the chromosphere structure are also acquired. Experimental results indicate that contour line on maps can identify the dark and sightless field structure. The more abundant atmosphere motion details are acquired on the sun or fixed stars under solar eclipse with low ISO and within limit exposure and un-PS picture. The shape of spray and spurting matter and the solar wind’s structures can be also found.

The way of development of modern space optical sensors is large aperture and long focus length. Long focus length means the distance between primary and secondary mirror will be longer, and tolerance of secondary mirror means more difficult to guarantee. For the secondary mirror strict tolerances, a design of space truss structure is presented. According to the theory of statically determinate spatial frameworks, two schemes are designed and were analyzed through finite element analysis software. We got the first three inherent frequencies, the angle around the X axis and the rigid body displacement along Y axis. The results show that the first mode of the scheme 1 is 78 Hz, the angle around the X axis is 2″, and the largest rigid body displacement along the Y axis is 0.04 mm. For scheme 2, the first mode is 113.7 Hz, the angle around the X axis is 1.5″, and the largest rigid body displacement along the Y axis is 0.022 mm. Two schemes can meet the design requirements, while scheme 2 is more excellent than scheme 1. The theory of statically determinate spatial frameworks provides a reliable theoretical basis for the structural design of space camera.

Using single radial immunodiffusion method to measure the Hemagglutinin Antigen (HA) content of influenza vaccines, the key is to determine the proliferation ring diameters. The traditional methods for the measurement of proliferation ring diameters use a vernier caliper or high-definition scans, which all have human disturbance that result in low measurement accuracy and stability. Based on machine vision technology, automatic measuring instrument for HA content is studied. The getting proliferation ring image by CCD camera and caculating the proliferation ring diameters by image processing, realize the operations automation and intelligence for acquisition of proliferation ring diameters data, calculation of HA content and analysis of results, which effectively improves the measurement accuracy, and eliminates the human disturbance. Experiments show that the measurement accuracy of this instrument is up to 2μg/mL, with relationship coefficient more than 0.99(R2>0.99), and high stabilization of less 1% coefficient of variation, that meets the measurement needs. The instrument is suitable for every kinds of HA content. The results are easy to store, transport, and easy to secondary treatment.

This paper reports a novel 3D (three dimensional) reconstruction method from optical microscopic images. The method can effectively reconstruct 3D microstructures based on digital image processing and precision mechanics, greater than the field of view of an optical microscope of 0.647 mm × 0.483 mm. Firstly, partial-sample images are taken on different focal planes for the whole microstructure. Those partial-sample images are merged to create a sequence of whole-sample images of different focal planes. The sequence of whole-sample images is spliced to keep only the area of interests by using a highly efficient image splicing methods. The spliced images sequence is then used in 3D reconstruction of the microstructure by a coincidence algorithm. Experimental results show that the method is simple and effective, and the 3D reconstructed image provides richer information of the microstructure.

A new design method for binary barcode was proposed, which was suitable for the photoelectric location coding of absolute style. The traditional photoelectric coder was mainly based on the technology of Moiré fringe grating, the coder was simply structured and could convert the displacement measurement into the electric impulse. However, the application of photoelectric coder was restricted because of the relative measurement style. The new method for the binary code of this dissertation, designed a grating which included three different types of frequency information in the single code channel, and had constructed a new binary barcode by the binary impulse encode modulation and method of the error diffusion. Part of the photoelectric location barcode was imaged on the CCD, and the code information was captured by the CCD simultaneously. By using of the FFT, filter and IFFT, we will found the phase information of the three frequency grating. By using the Excess Fraction Method (EFM), we can get the absolute phase distribution so we can get the information of the absolute position. The new method has preserved the advantage of the simple structure of the single code channel Moiré fringe grating. At the same time, we can let the absolute position measurement come true just like the multi-code channel grating coder. This paper introduces the detailed description of the coding method, the processing method for code information and experimental data of the computer simulation. The new method proposed has a broader application in absolute photoelectric position coder and the digital level.

Fizeau interferometer with the advantage of common optical path which the reference light and test light are passed becomes main structure of interferometer. The structure of interferometer and its influencing factors in theory have been analyzed by lots of literature. But how to simulate Fizeau interferometer is the problem we faced. This article first introduces the basic principle of Fizeau interferometer and phase unwrapping, and the phase unwrapping procedures based on Matlab were written. Then Zemax optical design software was used to model Fizeau interferometer and four-step phase shifting was used to simulate surface testing. By the phase unwrapping procedures, the true surface to be tested were got. The simulation of testing result is in line with the result given by real Fizeau interferometer. Simulation result shows that the model can simulate the actual optical interference.

In view of the problems of the current thresholding method based on 2-D minimum cross entropy, such as computing complexity, an improved image segmentation method based on 2-D minimum cross entropy is presented. Firstly, a neighborhood mask was selected according to the image type including the noise and a corresponding 2-D histogram was created to improve the segmentation performance. Then, the thresholding formula of 2-D minimum cross entropy was simplified, and through the defined array operations, the recursive algorithm was combined into a new search algorithm. The gray limits of the image and the neighborhood image were obtained, and in the limits, the new recursive algorithm was used to search the best threshold vector and to reduce the computing complexity. Finally, the neighborhood image was segmented with the key threshold to obtain better segmentation effects. Experimental results show that the proposed method’s segmentation effect and its anti-noise are better than those of the current thresholding method based on 2-D minimum cross entropy, and its computing time is much less, below 0.05 second.

The traditional camouflage is based on a number of irregular spots or stripes, whose edges are smooth as well as they have clear boundaries between different colors. The visual distinction degree is so strong that the camouflage effect is debased. The digital camouflage can break the boundary of a camouflaged target between different colors, and make it blurred and broken, so as to attain a better camouflaged effect. An algorithm which can generate digital camouflage figures is presented with background images of the target. It quantizes the colors of the background, extracts the dominant colors, and then determines the size of a mosaic block. Finally, the digital camouflage figure is generated. The experimental result shows that, the digital camouflage is easier to blend with the natural background than the traditional camouflage, and achieves a good effect of the optical camouflage.

In order to quantify background clutter reasonably, some collected images had been evaluated by using three metrics of background clutter. The metrics consisted of the probability-of-edge, the statistical-variance and the phase–correlation. The performance of these metrics was verified with the statistical results of target acquisition. It is demonstrated that the metric of probability-of-edge is more stable than others. Moreover, the fluctuation of the gray value in the image has a great effect on the performance of statistical-variance metric and it is not suitable to quantify the background clutter for the scene in which the mean of statistical variance of the background does not consist with the saliency of the target. And the homogeneous background has a great influence on the performance of phase-correlation metric and it is more suitable to quantify the background clutter for the scene in which much more information is contained.

Referring to the problems of Synthetic Aperture Radar (SAR) image registration, an approach of SAR image registration based on affine invariant Scale Invariant Feature Transform (SIFT) features is proposed. First, the affine invariant SIFT descriptors are constructed by improving the original SIFT descriptors and the control points are rough matched by using the improved SIFT descriptors. Then, a new proximity matrix is built according to the scale ratio and the difference of orientations among the rough matched points and the similarity of the intensity around the rough matched points. Finally, the exact matching points are determined by using Singular Value Decomposition (SVD) method, and the precise image registration is performed. Experimental results show that the proposed algorithm has a better performance than original SIFT algorithm and SIFT+SVD method, and the accuracy of our algorithm can achieve sub-pixel level.

A novel approach based on the Edge Improved Local Binary Pattern (EILBP) visual feature and PLSA model for scene classification was presented. Moreover, the EILBP features could not only capture the distribution information of the local edge gradient and direction, but also obtain the local structures information for describing image. At first, EILBP features were extracted from edge regions as visual words, and then these visual words were formed by clustering method. After that, the Bag-Of-Words (BOF) model was used to represent the image contents. At last, the potential semantics was excavated by PLSA model and the confusion matrix was obtained by KNN classifier. Experiment results show that this method achieves higher accuracies, especially performs well in the images with much edge contours and also it needn’t require experts to annotate the scene content in advance.

An iris location algorithm based on searching small-scale edge regions is proposed. In this algorithm, after analyzing the gray features and morphologic processing, the binary image can be obtained firstly. Some parts of the inner boundary are found by combining the edge image using sobel operator with the binary image. The regions of search are segmented by using canny operator and geometric characteristics of iris. Then, the proposed five characteristics are used to obtain some parts of the outer boundary. Finally, least square method is used for fitting the boundary. The experimental results show that the proposed iris location algorithm can locate the iris quickly and accurately, and it achieves good location for the iris images of different quality.

Curvelet is a multiscale and multidirectional image transformation tool, which can efficiently overcome the redundancy of wavelet in expressing the singular feature along curves of the image, and can obtain a sparse feature representation. Moreover, based on the consideration that high-dimensional image may exist in lower dimensional manifolds, manifold learning is performed on the Curvelet features so as to find low-dimensional structures, which is used for face recognition. Experiments show that the Curvelet features further processed by LLE show better clustering ability than the LLE. Compared with the already existing Gabor-based manifold learning, Curvelet-based manifold learning perform better under both facial expression and illumination changes, and either case sees valuable improvements. Experiments in the Essex expression and Yale B lighting face databases prove this point.

To recognize optical character with noise pollution rapidly and accurately, a novel approach for character feature extraction based on statistics and fuzzy membership is proposed. Compared with traditional method, this approach has a higher degree of differentiation in feature space increasing 33% of minimum inter-class distance. Applied in Radical Basis Function (RBF) neural network, under the influence of different font size and image background noise pollution, character recognition rate is up to 75%. Theoretical analysis and experimental results show that, compared with traditional methods, this approach achieves a better anti-noise performance, greater degree of differentiation and lower time and space complexity. It can be simpler, more comprehensive coverage characters' features with wide application. This approach has been applied to the actual system and achieves good results.