Traditional hyperspectral anomaly detectors background statistics are easily contaminated by anomalies and not robust that is difficult to detect anomalies of nonlinear mixed. In response to these problems, the Kernel feature projection theory is utilized, robustness analysis method is introduced in the construction of the anomaly detector background information, and a robust anomaly detection method is proposed. Using this method, c hyperspectral data from lowdimensional space can be mapped to high dimensional feature space without specific nonlinear mapping function, the background and anomaly targets can be expressed by a linear model, and a robust detector is constructed in the feature space. This method reveals highorder features between the objects on ground surface and can reflect complex spectral characteristics of target and surface features distribution. Experiments of real hyperspectral images and simulated data can prove: 1) the proposed anomaly detection method has a better receiver operating characteristic (ROC) curve and area under the curve (AUC) statistics and has a greater degree of separation of the target and background; 2) in kernel feature space, exclusion of anomalies contamination on statistics of the background improve the detection accuracy; 3) feature extraction can make better utilizing spectral diversity distinguish anomalies and background, which is an important step of anomaly detector.

The imaging principle and the structure of a Hadamard transform spectral imager with a digital micromirror device are presented. A phenomenon of energy excursion which is caused by mismatching of the mask unit size and the detector pixel size is expatiated. To fix the recovery images, a matrix correction algorithm is proposed. The 7order images of plants acquired by the Hadamard transform spectral imager are used and the correction algorithm is operated on the recovery images with error, which reduces the phenomenon of dark stripes effectively. In order to improve the image quality of the recovery images, two dots in and off two adjacent dark stripes are selected to show the spectrum curves. The results show the close spectrum curves of the dot off dark stripes and the different curves of the dot in dark stripes. The comparison illustrates the validity of the correction algorithm.

Compared to conventional imaging spectrometry, computational imaging spectrometry has the advantages of high throughput snapshot imaging etc. But because of the presence of the dispersive element, computational imaging spectrometry suffers the effect of spectral line bending. To study the effect of spectral line bending on signal acquisition, spatialspectral aliasing and reconstructed result in computational imaging spectrometry, combined with the principle of computational imaging spectrometry and reconstruction algorithm, the relative peak signaltonoise ratio of the reconstructed image and the maximal error of the reconstructed spectral curve with different spectral offset were calculated and analyzed. The experimental result showed that spectral offset of the signal acquired by the detector will change the degree of spatialspectral aliasing. The reconstructed results with spectral line bending exhibit distinct errors compared with no spectral line bending. And both sides of the reconstructed spectral curve tend to smooth. In order to reconstruct the object scene with high accuracy, spectral offset should be no more than half a pixel for computational imaging spectrometry with 10 nm resolution.

In order to overcome the shortcoming that the luminous flux of the classical spectrometer is small due to the occlusion of the entrance slit, a design concept of the new kind of Hadamard transform spectrometer is proposed. A method of shaping the beam by a cylindrical lens is given. Using theoretical calculations of the spectral resolution ability and the pixel resolution ability of the spectrometer, containing a digital micromirror array, he limit class of the spectrometer resolution can be determined. Based on the geometric theory of optical imaging, an optimization program of the more focal length of the imaging lens is proposed, which is able to obtain the higher spectral resolution and the more beneficial to the mechanical alignment. The advantages of the new Hadamard transform spectrometer are that it has a high luminous flux, the signaltonoise ratio is 5 times more than that of traditions, and the smaller spectral bandwidth is 3.5 nm, providing convenient conditions for the detection of weak spectral signal. In addition, the nearinfrared detector selected by the spectrometer needs no refrigeration, which will cut the production costs of the Hadamard transform spectrometer, with a stronger market competitiveness.

Usual analytical methods are complex and require rigorous experimental condition. In order to meet the demand of fire product supervisory department in rapidly detecting the quality of dry chemical, a rapid and nondestructive method was proposed to determine the moisture content in ABC dry chemical using near infrared spectroscopy of fiber diffuse reflection. The calibration models were established for determination of moisture content of ABC dry chemical by partial least squares and 47 ABC dry chemical sample spectra. Spectral data preprocessing and Spectral region selection were discussed. Correlation Coefficient of Calibration Set of the model was 0.976, root mean square of error cross validation was 0.037 1, correlation coefficient of prediction set was 0.952, and root mean square error of prediction was 0.021 6 in the predicted rang of 0.075%~0.334% for moisture content of ABC dry chemical. Using the Chisquare test method, repeatability standard deviation measurements all belonged to the same overall at a 95% confidence level. All the relative standard deviations were less than 7.0%. The near infrared spectroscopy method is used to test the moisture content of ABC dry chemical is feasible.

The compressive sampling spectral modulated technique is developed in order to avoid time penalty and promote throughput of spectral imaging system, and a digitalmicromirrordevice based compressive sampling multispectral imaging system is designed. In the technique based compressive sampling spectral information is sampled below Nyquistcriterion with coded aperture to reduce spectral data greatly. In the experiment, the cubic spatial/spectral data cubes of the 612 nm laser and colored striped fabric are reconstructed from 2Dimension images obtained from detector that contain spectral information using double shrinkage fast iterative algorithm. The results indicate that a digitalmicromirrordevice based compressive sampling multispectral imaging system has images with highthroughput and high resolution, can compress the spectral simultaneously, and the compression ratio can reach 31∶1.

A 3D edge surface denoised tracking algorithm is proposed to reconstruct high accuracy edge surfaces from the noisy industrial CT slices based on the 3D fractionalorder integral. The 2D fractionalorder integral method has effective denoising ability to preserves the texture detail of the image, and it has low computation complexity and easy implementation due to the filtering mask. In this paper, the 2D fractionalorder integral has been extended to threedimensional images, its 3D continuous theory and the discrete filtering masks are also proposed, we call it volumetric fractionalorder integral. Since the Laplacian operator shows the sensitivity to the noise, the traditional 3D edge surface tacking method cannot extract the high precision 3D edge surface from noisy slice images effectively, the 3D fractionalorder integral is added to the tracking method to overcome the existed shortcoming. Our method is able to detect and extract the 3D edge surface of subvoxel accuracy from the 2D noisy industrial CT slice images. The experiments have reported very encouraging results according to signal noise ratio and visual effect by comparing it to the tacking method based on 3D Gaussian denosing method.

To overcome issues in the conventional parallel threedimensional imaging system, such as the smaller threedimensional public scene, none of the horizontal positive parallax resulting in threedimensional reproduction only in front of the screen and causing visual fatigue easily, an improved parallel 3D imaging method was proposed. Based on the portion of the readout pixel array reserved but not used for display in most image sensors, the display array was made a corresponding horizontal offset by reference from offaxis parallel display technology and then fed in 3D display monitor. Through the analysis and derivation in the established mathematical model, the horizontal positive parallax production was realized, so threedimensional reconstruction could be felt behind the screen, and the stereo visual effect was greatly enhanced. The proposed method was proved feasible on the selfdeveloped system of 640×480@120 Hz with selection of the CCD resolutions to 752×582, and the angle of threedimensional public scene could be increased from the conventional 29° to a maximum of 34°. The proposed method provides a new direction for the appropriate design of the positive and negative parallax and for the study on the improvement of the stereo effect.

In order to address the spectrum fragmentation problem in flexible grid optical networks, the spectrum fusion oriented routing and spectrum allocation algorithm and spectrum defragmentation algorithm are proposed. The notions of spectrum allocation window, spectrum allocation point and spectrum fusion window are introduced to optimize the network resource utilization. To retain the spectrum consecutiveness as much as possible when establishing a lightpath and reduce the spectrum fragments, the proposed RSA algorithm considers the spectrum utilization in candidate links and their adjacent links. The spectrum defragmentation algorithm is used to make enough room for the blocked request by rerouting already established lightpaths. Simulation results reveal that the algorithms significantly improve the blocking probability performance and resource utilization under dynamic traffic in flexible grid optical networks.

In order to reduce the performance impairments of directdetection optical orthogonal frequencydivision multiplexing (DDOOFDM) system caused by the nonlinear effects of electroopitc modulator and fiber chromatic dispersion, the impairment mechanisms are analyzed theoretically. Based on the analysis results, a digital preprocessing technique which combined predistortion and preemphasis methods to combat the nonlinearity and power fading, respectively, is proposed to employ to OFDM transmitter to improve the system's performance. The proposed method is verified by numerical simulation. The simulation results show that error vector magnitude (EVM) of the received signal can obtain a 2 dB improvement when both modulation indexes are optimum, compared to without employing the predistortion technique at the optical backtoback transmission case. With the increasing of optical fiber transmission distance, the power of high freqency subcarriers of the received OFDM signal is attenuated severely, then the BER performace of the system is deteriorated dramatically. The sensitivity of OFDM receiver can be improved significantly with the proposed digital preprocessing technique with low computational complexity.

In order to obtain the whole lens′ precision surface information and reduce its errors transfer and accumulation, a new global optimization stitching algorithm based on weight coefficient is presented. The common region of the two neighborhood subaperture can reach the optimal match and the stitching error is minimized by using this algorithm. The multiapertures stitching simulation is carried out, and an actual experiment is carried to the flat lens of 150 mm. A new subaperture position method based on the image edge contour feature extraction is introduced and the error factors are analyzed for stitching. Both the simulation and the experiment results show that this global optimization stitching algorithm is good for reducing the transfer and accumulation error, which exist in the traditional method, and realizing the high precision subaperture stitching measurement.

Space and airborne remote sensing urgently require large relative aperture and wide fieldofview imaging spectrometer. Its telescope must have large relative aperture, wide field of view, telecentric image space, and high image quality. Based on the geometric imaging theory of commonaxis threemirror systems, the design problem of large relative aperture, wide fieldofview, telecentric offaxis threemirror anastigmatic (TMA) telescope are studied. The procedure for calculating initial structural parameter is programmed. By means of field of view offaxis, an telecentric offaxis threemirror telescope system is designed. The spectral range is 200～1 000 nm, the focal length is 210 mm, the relative aperture is 1∶2.5, and the field of view is 14°. The primary mirror and third mirror are sixorder aspheric surfaces, and the second mirror is a quadratic surface. RMS diameter of the spot diagram is less than 16 μm, 80% of the enclosed energy is in a pixel, the MTF is more than 0.75 at Nyquist spatial frequency 22.2 lp/mm, the distortion is less than 0.2%, and each specification satisfy the application requirement.

An improved machine vision illumination system was proposed that used general LED light sources. General LED light source with the package type of surface mounted device, or chip on board was used in the new system. A uniform light system was used to make sure the illumination uniformity when the new system switched different light colors of general lighting LED light source. Based on the edge light principle of the nonimaging theory, a method of simultaneous multiple surfaces was used to design the universally uniform optical system. The system on the premise of maintaining the optical system being fixed can switch multiple light sources without specially designed uniform light source. It ensures the uniformity of target surface illumination and expands the illumination subsystem′s light color range. The comparison between several groups of Monte Carlo ray tracing simulation shows that the general illumination light sources meeting the requirements of sizes can be integrated into the system as long as the light source used in the new system meets the design requirements of sizes; the nonuniformity of the illuminated area will not lead to the decrease of distinguishing degree; so the proposed light source is suitable for the machine vision system that requires not too high energy efficiency and high uniformity of illuminance of the target area.

The photonic crystals (PCs) with defect layer containing negativeindex materials coated over a SiC substrate are investigated by a transfer matrix method. The results show that this coherent thermal radiation source has very sharp emissivity peaks within a narrow wavelength band at a welldefined direction; the quality factor of the emissivity peak is high. This emissivity peak is not sensitive to the emission angle, polarization, and scaling factor of the photonic crystals in the structure. The value, quality factor, and position of the emissivity peak can be effectively controlled by tuning the thickness and refraction index of the defect layer.

Basic structure of GiresTournois mirror is described and the dispersion performance is calculated. The factors affecting the performance of the GiresTournois mirrors are discussed. The results show that the layer number of high reflector affects the reflectance of the GiresTournois mirrors but the thickness of the GiresTournois cavity and the layer number of the top reflector affect the dispersion performance of the GiresTournois mirrors; to achieve good design performance, the layer number of high reflector, the thickness of the GiresTournois cavity and the layer number of the top reflector are selected to be 40~60, λ/2 or λ and less than 5.

In order to obtain the information of elevation which threedimensional reconstruction needs in the spotlight mode inverse synthetic aperture ladar imaging, a target reconstruction algorithm was designed based on spotlight mode inverse synthetic aperture imaging ladar groups. The systems were placed inverse synthetic aperture ladar in the three vertices of an equilateral triangle. According to the relationship between the position and the inclination angle, the elevation information for reconstructing threedimensional target could be provided by another two ladars. The elevation functions of three inverse synthetic aperture ladars were calculated when the measured target flied into detects area of the system in any direction. Simulation result shows that the speed of the measured target has no significant effect on the elevation function. Angular relationships of inverse synthetic aperture ladar and target have a significant contribution to the changes in elevation information. In the process of information fusion for different position elevation, the selection of distributed manner for inverse synthetic aperture ladar impacted the utilization of echo data. Because different angles change produces continuous changes in elevation information, this method can effectively reconstruct target threedimensional image by elevation information obtained.

The output characteristics of graphenebased passively Qswitched pulses at 2 μm wavelengths were reported. On the basis of linear resonant cavity, Tm3+doped fiber and graphene saturable absorber mirror served as gain medium and passively Qswitched device respectively. Tm3+doped fiber was end pumped by 792 nm semiconductor laser. Light in the cavity was focused onto graphene membrane through a set of collimation and focusing lens. Passively Qswitched pulses centered at 1 958 nm wavelength were acquired successfully. The minimum pulse width of 1.02 μs was obtained when pump power rise to 3.0 W, and corresponding average output power of 26 mW, repetition rate of 116 kHz, single pulse energy of 224 nJ were obtained respectively. Furthermore, average output power and pulse width presented approximate linear relationship with incident pump power. The experimental results indicate that excellent saturated absorption characteristic of graphene is effective for passively Qswitched pulse operation at 2 μm wavelengths.

According to the characteristics of new laser fuze optical transmitter which emits 90°×2° linear infrared laser beam, measurement technique on transmitter optic component is researched. New laser fuze optical transmitter test system is developed, and a method based on two linear array CCD scanning imaging is proposed. The twolinear array CCDs are placed at Z1, Z2 positions and are separated by an angle Φ. Near field and far field light beam distribution are acquired to realize the automatic measurement of parameters such as the view direction relative intensity distribution, beam divergence angle, pitch angle and optical power by beam analysis software. Experimental measurement results indicate that the parameter measurement values are in the acceptable range with measurement accuracy of 1′, repeatability precision of ≤5%. The system performance is well in repetition, dependability and automatization.

Aiming at the existent burst point time measurement device was unable to show the burst point time measurement situation and had a big measurement error, avalanche photodiodes (APD) were used as photoelectric detectors, wireless data transmission modules were applied to transmit burst signal timely and directly timed by PC instead of traditional timing circuits to improve timing accuracy. The amplifying circuit based on APD was designed, photoelectric receiver module was built, and the falling edge width of the output pulse signal was less than 25 ns. Of the wireless data transmission module, the transmitter was contacted to photoelectric receiver module, and the receiver connected with PC. Through the wireless communication between the transmitter and the receiver, the pulse signal when the ammo was outing of the muzzle and exploding was passed on to the PC that was 1~3 km away. The measurement situation was monitored realtimely, the burst point time was read intuitively, and interference signals could be artificially filtered. Measuring the burst point time of a test bomb, which was 124.056 ms, the results of the test experiment shows that the device is able to effectively realize remote realtime monitor the burst point time with high measuring accuracy. This design can provide a reference for the optimization and improvement of the burst point time measuring device.

The basic principles based on dynamic characteristics of laser speckle for monitoring of the paint drying process was studied. The processing algorithms of realtime detection were analyzed.A differenceaverage algorithm of dynamic speckle was proposed. In this algorithm, gray distribution histogram of the difference image was used to qualitatively describe the strength of the dynamic characteristics of the laser speckle, and the average gray level of the difference image was used to quantitative describes the changes of speckle. Experimental results show that dynamic characteristics of speckle can be used to nondestructive and online monitor the paint drying process, and the differenceaverage algorithm can be given realtime results in the monitoring process with shorter computation time and lower hardware requirements.

The result of target recognition is always influenced by clutter background, low contrast and deficiency luminance of image. Besides the factors mentioned above, detection and recognition of partial masked target has the character of imperfection and information loss of target, which leads to recognition problem. All the harmful factors cannot make target recognition be realized by hybrid optoelectronic joint transform correlator due to the weak correlation peaks or even no peaks. Aiming at this phenomenon, partial masked target image in hybrid optoelectronic joint transform correlator is processed by the method of adaptive threshold, and power spectrum is processed by Mexican wavelet transform. The combination depresses the noise disturbance effectively and enhances the energy of diffractive light of obscuration target and reference template. This enhances the contrast and energy of correlation peaks of partial masked target and improves the ability of detection and recognition for partial masked target with joint transform correlator. The experimental results show that the combination of adaptive threshold and Mexican wavelet transform can improve the ability of detection and recognition for partial masked target with hybrid optoelectronic joint transform correlator effectively.

Current paper defects identification methods has two radical problems. First, every current method can only identify one or few defects. Second, current methods can hardly detect the complex paper defects accurately. In view of these problems, based on comprehensive analysis of paper defect features, research and summary of all kinds of paper defect identification methods, Fuzzy fusion device is used to conduct feature layer fusion with some paper defect characteristic values, and combine multiple paper defect identification methods, aiming to achieve more efficient and comprehensive paper defect identification. According to the construction equivalence between RBF Neural Network and fuzzy reasoning, the paper defect features information fusion system on basis of RBFNN has the advantage of simple structure and rapidity. Experiments have shown that the method presented is practicable to identify the primary paper defects accurately, including complex paper defects.