In order to solve the problem of passive tracking low-altitude maneuvering target with acoustic network, a Digraph Switch Variable Structure Multiple Model (DS-VSMM) algorithm is proposed. First, acoustic sensors are networked and their time-delay bearing data is aligned, which improves the fusion performance of acoustic network. Second, a variable structure multi-model algorithm is used to estimate target state, in which the weight of new activated model is assigned by digraph switch method. Third, the system output is fused by multiple parallel models. Compared with the IMM method in simulation, the proposed algorithm can match actual target trajectory with less computational complexity and better accuracy.

In order to analyze the combination of high conflict information of moving target tracking in complicated environment, a modified combination rule of evidence by distributing local conflict to local focal element is proposed in the framework of Dezert Smarandache Theory (DSmT). Based on thorough theoretic analysis on the relations of conflict and correlative factors, location and color cues of moving targets were combined by the introduced method, and the tracking model of combined multi-source information was established. Experimental results show that the target tracking is realized under occlusion conditions by the method, the efficiency and reliability of multiple video moving target tracking are improved greatly, and the robust tracking results can be obtained, which has high evidence conflict.

Classical multi-channel algorithms usually employ minority discrete channels. It is considered that the channels are independent and their multi-mountain like distribution is neglected. Double-layer particle filter for target tracking with multi-channel energy extreme points set method is proposed. Visual target is firstly decomposed to multiple frequency and angle channels. And then energy extreme points set is calculated as target feature. Double-layer particle filter algorithm framework is established. Target motion state is globally constrained by first-order particle. Second-order particles represent the accessory multi-channel features and cluster to feature peak by local optimization method. Multi-channel features are examined and then second-order particles are weighted. The result is feedback to first-order particles as weighting parameters. Thus global constraint and multi-feature reliability distribution are balanced. Experimental results show that the algorithm runs steadily with moderate computational complexity. Compared to correlation methods, the average tracking error is decreased by about 0.5 pixels.

To improve the performance of Converted Measurement Kalman Filter (CMKF), a more accurate evaluation method for converted measurement error statistics (means and covariance) was presented. There were two sets of information that could be used to evaluate the converted measurement error statistics, i.e., measurements of the target and a priori state estimate of the filter. Firstly, the proposed procedure in this work derived the means and covariance of the filter’s a priori spherical state estimation errors. Secondly, a more accurate evaluation of the converted measurement error statistics was obtained according to the a priori spherical state estimate of the filter instead of measurements. Finally, the improved evaluation method for converted measurement error statistics was utilized to implement the CMKF algorithm for a target tracking scenario. The simulation results show that the proposed method can provide superior performance in terms of convergence and estimation accuracy, especially in the case of significant measurement noises.

Because motion estimation algorithm occupies 60~80% of the entire computation in image stabilization algorithm, and the difficulty increases with the improvement of image resolution and real-time stability of continuous image sequences, a GPU-based phase correction strategy for stabilization phase correction algorithm is presented. Based on mathematical models, a unified parallel architecture of image processor and pixel single-threaded mode are used, the phase of motion estimation overlapping composite structure is designed, improving the performance of the parallel thread synchronization. The computation time of parallel FFT is reduced. At the same time, change the scale of parallel threads, adjust the graphics processor storage, improve memory access performance data and motion estimation phase of concurrency, and estimate a stable rotation and translation etc. required for video in order to achieve fast and stable gray scale resolution of 1 k × 1 k purpose of continuous video. Experimental results show that the average time is about one frame stable 10ms, which effectively enhance the efficiency of video stabilization, and the possibility of image stabilization phase engineering applications.

To clarify the effect of sensor allocation form and the number of sensors on the localization precision, the localization performance of sample mean in multi-sensor triangulation localization is analyzed in the global coordinate system. First, the correlation between intersections and the Chebyshev condition is studied according to the consistency definition, which indicates that the sample mean in multi-sensor triangulation localization is a kind of inconsistency estimator. Second, combined with the effect of sensor allocation form on the localization precision, the approximate consistency condition of sample mean is analyzed. Finally, the relationship between consistency and multi-sensor localization precision, and the essential difference about consistency in the regular and irregular sensor allocation form are discussed respectively. Simulation results verify the analysis about the localization performance of the sample mean, which can be applied to multi-sensor passive localization algorithms based on sensor management.

Object tracking by using single feature results in a poor performance in complex scene, and the traditional template update strategy is not robust to target appearance changes. Therefore, the paper presents a multi-feature fusion tracking method by mean-shift and particle filter tracking framework. The particles were optimized by mean-shift thus overcoming the degeneracy problem. According to the distribution of particles, the fusion coefficient of each feature is estimated online. Besides, a dynamic layered-weighting strategy is used to adjust the updating weights adaptively according to the variety of each component in template. The experimental results show that our algorithm can deal with dynamic appearance changes of target, thus it is suitable for tracking in dynamic scene.

To improve the tracking performance of the O-E high speed tracking system, it is necessary to put forward the requirement of mechanical properties of the controlled plant according to system’s control accuracy index, which can be used to do mechanical working to make sure that the design can be realized. Specifically, that is to say the structure of the motor and reflection mirror’s transfer function is known, but the parameters keep unknown. Aiming at the request of horizontal high speed and accuracy measurement of the O-E high speed tracking system, first, the system’s mathematic model has been established. Second, some strategies such as trigger guiding, two closed loop control as well as compound control have been presented and applied into the system. With bisection method, the parameters of the motor and reflect mirror have been adjusted. Third, the requirement of the mechanic structure parameter as well as the load resonance frequency have been confirmed to make sure the system can meet the tracking accuracy. Experimental results indicate that the maximum tracking error can be limited to less than 0.086° and the regulator time can be 0.04 s which manifests that the design can satisfy the system requirements of swiftness and high accuracy, and the design also ensures the realization of the control.

In the deep space, targets are far away from the cameras and become dim targets with less shape and gray features in the image planes. It is an essential problem of target correspondence across multiple cameras. A novel algorithm based on geometries consisting of detected targets in multiple cameras was proposed. The algorithm firstly compensated the rotation effect of the detected targets set. Then, Delaunay triangulation was used to divide set of compensated points. Finally, triangle similarity was employed for target correspondence. According to simulation and analysis, the algorithm dealt with target correspondence across multiple cameras in deep space effectively.

To solve the problem of the track initiation typically encountered in the presence of clutters, missed detections and unknown number of targets, a track initiation algorithm for a passive sensor system is presented. First, a part of false intersection points is first eliminated by some heuristic rules, then the observation space is divided into uniform-size grid by using probabilistic grid concept, and the probability of the measurements falling within each grid is calculated. Finally, the tracks are detected by using the serial Hough. Simulation results show the presented algorithm can initiate tracks effectively. Besides, the algorithm reduces the ratio of false track initiation.

The traditional object tracking methods are difficult to deal with the interference of complex background, the variety of object shape and the irregular change of object position. A new object tracking method based on the weighted Lucas-Kanade algorithm is proposed. Firstly, a reasonable estimation of object position in current frame is obtained according to the search template. This position is used as initial iteration parameter of the weighted Lucas-Kanade algorithm. Secondly, the weights function is calculated and the accurate object position is achieved by tracking the current template and the initial template in current frame. Finally, the template update strategies under the complex background and the object shape variety are implemented. Experimental results based on a large amount of measured data show that the proposed method can effectively realize stable tracking of object in complex background.

According to the structural design scheme for the main tube of a tracking system fixed on a moving vehicle, the Finite Element Method (FEA) model is established with the software Ansys Workbench. The first ten natural frequency and the distributions of the structural strength are obtained. And it is concluded that the relative position of the primary and secondary mirror is the interesting factor. Therefore, three parameters are defined to describe the variance of the relative position. Finally, the transient responses of three parameters due to the load impact are analyzed. The analysis results are provided for the dynamic optimal design and the vibration analysis of the whole tracking system.

The method of measuring absolute distance with the optical frequency comb of a femtosecond laser was proposed. A femtosecond laser was used as the light source of performing absolute distance measurement. A Michelson type interferometer structure was built. Exploiting the principle of the dispersive interference, the spectrum was analyzed. Then, the phase difference which was caused by the optical path difference of optical interference was obtained. Finally, the optical path length difference was measured. The experiment result shows that our length measurement method has high accuracy with the resolution of nanometer level. The minimum measurable distance is 9 μm, and the non-ambiguity range of length measurement is 5.75 mm. Compared to the limited measuring range of traditional white light interferometry, the maximum distance can be extended to arbitrary length.

The profilometry based on the fringe contrast is a vertical 3D measurement technique, which can measure the complex objects with acute changes in their surfaces. A modified contrast measurement profilometry based on the previous method is proposed. In this method, sinusoidal fringes are projected on the surface of the testing object laid in front of imaging plane of the fringe and captured through a beam splitter by a CCD camera. We can obtain the fringe contrast using phase-shifting technique, and then restore the height of the object through the map of the contrast and the height. In the measurement process, we only need to capture fringe patterns in one position, without changing the image plane in several times, so it is faster and more convenient than the existing method. The experimental results prove this method can restore the height of the complex object effectively.

The number of fabric defect classes is large and a single method is only effective for specific types of defects. In order to detect defects more effectively, a novel fabric defect detection algorithm is presented based on combining wavelet multi-scale product and mathematical morphology. After the dyadic wavelet transform on the images, mathematical morphology operations are utilized to the low frequency sub-image to obtain defect shape features. Then, the wavelet multi-scale product methods are used to the high frequency sub-images to realize suppressing noise and enhance defect edge linear features. Finally, the results above are fused to get the final result. Experimental results show that the false alarm rate and computational time are low, and the detection rate is high. The comprehensive performance is superior to the Gabor and wavelet algorithms.

In order to solve the conflict between noise suppression and detail signal reservation of Near Infrared (NIR) spectroscopy, a wavelet domain noise reduction method is proposed based on the noise variance estimation. The method establishes two states Gaussian Mixture Models (GMM) of high-frequency coefficients in wavelet domain, uses Expectation Maximum (EM) algorithm to figure out the factor of model, proves that the model can accurately estimate noise variance, and puts the model to wavelet threshold noise suppression. By establishing the wine alcohol Partial Least Squares (PLS) prediction model of NIR spectroscopy, and comparing Penalty threshold, Brige-Massart threshold and the default threshold wavelet threshold denoising effect, the experiment validates that the method has better noise reduction effect than other normal method, which improves the stability of NIR model.

A portable infrared detection system for electronic circuit fault based on ARM11 is introduced. The thermal image of circuit board to be detected is captured by CCD camera. Then, the image format of the thermal image is transformed by TVP5150, which is used as a video decoder chip. The infrared image will be transported to core processor S3C6410, and the S3C6410 will deal with the thermal image by image processing algorithms. At last, the thermal image of circuit board to be detected will be compared with the thermal image of correct circuit board, and the differences between the two thermal images could come out. The results of experiment show that the detection system has characteristics of high accuracy, high precision and high stability. At the same time, system size is reduced, which makes the system have the better portability. The system could be applied for detecting electronic circuit fault non-intrusively, which make the system have the higher safety.

The temperature testing of bridge wire radiation has many difficult technical problems, for instance, the small size of detected target, the small volume of infrared radiation, the large signal loss in distant transportation, etc. A two-wavelength fiber optic temperature measuring system was developed. The principle of the thermal radiation about the Electric initiating Explosive Device (EED) is analyzed and two-wavelength mathematics model is derived. Effective optical fiber coupling combined with methods for developing SNR in system are adopted. The InSb (response range is 0.8～3 μm) and HgCdTe (response range is 3～5 μm) are chosen as the two-wavelength detector. Inductive current and temperature measurement are realized using uncontacted method to EED target. The test device has strong electromagnetic shielding ability, and is free of environmental temperature effect. The measurement error caused by the temperature drift is avoided. The measurement precision is 1 K. It becomes the base of achieving the temperature measurement further in extremely weak thermal radiation of EED like bridge wire.

A major drawback of the Contourlet transform is frequency aliasing, which makes the basis images not localized in its frequency domain. A new algorithm is designed based on non-aliasing Contourlet after we analyze the cause of frequency aliasing. It is used to reconstruct a super-resolution image. First, an anti-aliasing pyramid is used by the new algorithm instead of the Laplacian pyramid. Then, the bi-cubic interpolation is used to do similar transformation in the high frequency domains, according to the similarity among the high frequency domains with different scales and different directions. At last, a higher-resolution image is obtained by inverse transform. Experimental results show the algorithm can get higher PSNR and better image reconstruction effect. It also avoids the distortion and blur caused by frequency aliasing compared with the algorithm based on the traditional Contourlet.

At present, multi-resolution analysis has become a popular method in pixel-level image fusion. However, those algorithms based on multi-resolution analysis with decomposition and reconstruction transform process is too complex and not suitable for real-time image sequence fusion. In order to meet the requirements of real-time fusion system, a novel Fast Mutual Modulation Fusion (FMMF) algorithm for multi-sensor images is proposed. First, the two source images were magnified by factors that were derived from the ratio of the corresponding pixel energy respectively. Then, an offset entry obtained by computing statistical parameters of source images is added to it. Finally, after the previous results are multiplied and normalized, the fused image is obtained. Fusion process consists of the addition and multiplication, which is a nonlinear combination process. Experimental results show that FMMF algorithm is simple and fast and its performance and efficiency are superior to those based on pyramid and wavelet.

A high-speed real-time image processing system based on System on a Programmable Chip (SOPC) of Field Programmable Gate Array (FPGA) was designed against the real-time requirements of complex algorithms for high-speed digital image signal and the existing problems that the system was updated in real time. The customized and efficient operating system Xilkernel was transplanted. And the methods of multi-thread programming, software simulation and real-time debugging online were applied. The real-time processing function of high-speed system has been achieved successfully. The operation of FPGA could be performed in parallel and the responsive time of its hardware could be accurate to nanosecond (ns) level. So the real-time processing properties of the system have more advantages over other processing platforms, and the real-time updates of the system without changing the hardware circuit have also been implemented.

In order to improve the visual effect of low illumination video, the high-quality daytime information of the same scene is often introduced to enhance nighttime video. A novel frame illumination-based method of video enhancement is proposed. In order to require daytime background, we firstly use averaging K frames method. Then, illumination of daytime background and nighttime video frames is extracted by using Retinex theory to enhance nighttime motion objects. Moreover, nighttime video frame’s motion objects are extracted by using different frames method. Lastly, daytime illumination is used to fuse nighttime video problem of the same scene between daytime and nighttime videos. For improving the perception qualities of the moving objects, we further propose Gaussian filter method for nighttime video enhancement which is efficiently and smoothly moving objects to the reasonable enhancement results. Experimental results show the effectiveness of the proposed algorithm can improve visual quality over conventional methods.

An approach of unknown phase shifts extraction and complex wave reconstruction in generalized phase-shifting interferometry is proposed by the combination of least square methods and generalized four-step formulae. A much-simplified algorithm is suggested to extract reference phase and then reconstruct object wave-front by replacing the conventional least-square method in above process with our specially derived wave reconstruction formulae for generalized phase-shifting interferometry. In this method, only one least-square equation is needed for one iteration to retrieve the reference phase, and the object phase can be calculated simultaneously. The feasibility and effectiveness of this algorithm have been verified by computer simulations. The results show that this innovation can keep the phase extraction errors mostly in 0.004 rad and save computing time by about 10 times compared with previous four-step LSM algorithm while it still ensures its high precision. Optical experimental results are also satisfactory.

With the development of the space imaging, all of the countries have increased the research work about the preparation and application of the new optical material. Because of the good thermal stability, high specific stiffness, low density and being easy to light weight, the SiC material is becoming the preferred material for the main optical components of the space camera. The paper describes a new type of SiC optical material which is used in the optical system mirror. And the common preparation technologies and related applications of the SiC are introduced too. The SiC material’s preparations and processing technologies are summarized seriously. The author focused on the differences of the SiC material and conventional glass processing, and compared the processing situation home and abroad. From the current situation of the SiC processing, the technology of Computer-controlled Optical Surfacing (CCOS) has enhanced the efficiency and precision of the SiC greatly, which has broad application prospects.