A face recognition method is proposed based on multi features description and local fusion decision. Firstly, we use Independent Component Analysis (ICA) to construct the global complementary subspace to roughly classify the test samples. Then the texture descriptor algorithms under three different definitions are used to construct local complementary subspace to obtain the posterior probability of sample which is difficult to classify by rough classification. Finally, we get the precise classification result of test sample on the local complementary subspace through setting grade scores based on the value of the posterior probability. The experimental results on ORL, Yale and FERET face database show that the proposed method better describes characteristics of the image and has lower time complexity and higher recognition rate. Compared with other methods, it also proves its effectiveness on the face recognition.

Aiming at the problem of large range focus in micro-vision system, a new focus search strategy was proposed. The improved SUSAN operator and wavelet transform operator were combined as a new focus evaluation function, which has the advantages of single peak and steep peak on both sides of the peak that was divided into a quick search of the flat area and Gaussian fitting steep zone. The effectiveness of the search strategy was verified based on the self-developed micro-vision system. Then, the motor was driven directly to reach the focal plane according to the fitting results. The experimental results show that the new focus search strategy has better effect on the aspects of the real time and accuracy.

In order to realize the automatic detection of the grating surface defect, a grating surface defect detection system based on machine vision is developed. First, the hardware platform of the system is built, so that key detection difficulties are analyzed and solved. Then, on the basis of the analysis and comparison of several traditional edge detection algorithms, the edge drawing detection algorithm is introduced to detect the grating surface image. Finally, according to the characteristics of the surface defects of various gratings, a number of threshold values are selected to define the surface defects of a variety of gratings and defects were extracted and labeled, and the detection algorithm is developed. Experimental results indicate that the detection accuracy of the system reaching micron level can detect the surface defects of many kinds of gratings in real time, with the detection efficiency up to 30 seconds, the recognition rate more than 94% and the mistake examining rate lower than 4%.

In order to achieve ocean optics ultraviolet remote sensing, it is needed to make an accurate radiometric calibration and system performance measurement for the UV imaging detection system. The method of integrating sphere spectral radiometric calibration, with xenon lamp as the light source, is used to calibrate the ultraviolet imaging detection system, which is designed and developed by ourselves. It’s verified that the output response of the UV detection system to the output spectral radiance from integrating sphere is an excellent linear relationship under the same integration time by taking the average of multiple samples, whose linear correlation reaches 99.99%. We have found the xenon lamp with a property of fluctuation, whose output spectral radiometric has ups and downs. In this paper, we firstly propose a model for xenon lamp, which is based on the relationship of three kinds of standard deviations of experimental results, UV imaging system and xenon lamp. This model can be used to correct the 48% inaccuracy of negative effect by the fluctuation of xenon lamp, and finally fix and get the real noise and SNR of whole system. Through the theoretical analysis, the establishment and verification of fixed model, it could provide theoretical and data support for xenon lamp integrating sphere UV calibration and parameters measurement like SNR.

In view of the traditional measuring method based on the triangle principle strict requirements on the structure parameters and difficult alignment, a new measurement model based on the space geometry principle was put forward. The corresponding plane equation of the two incline light screens calibrated was carried out by the least square method. The projectile flying velocity and the impacting coordinates would be calculated by solving the simultaneous equations among the ballistic trajectory line equation and the four light screens’ equations. The measurement method reduced the demands and simplified difficulty of the installation and adjustment. The method was verified by the real fire tests. The results show that the planar equation model and algorithm can be used successfully. Above all, the model and algorithm simplified the installation and adjustment of the four light screen array structure saved manpower and improved the measurement accuracy.

In structured light vision measurement, since non-uniformity of brightness of the common projector caused nonlinear distortion Gamma value distribution difference resulting in phase error problem, a self-adaptive subregional Gamma value pre-coding correction method was proposed. First, the principle of structured light stereo vision measurement and standard N step phase shifting method is introduced. Second, the projector and the camera's nonlinear distortion of sine grating and the Gamma value pre-coding correction method are analyzed. Then, on the basis of Gamma value distribution, according to the relationship between the change of Gamma value and the phase error, subregional threshold was calculated. Finally, according to the threshold, the measurement field was divided and subregional Gamma pre-coding correction method was implemented, and actual measurement was carried out on a standard white plate and a fender. Experimental results indicate, under the same conditions, compared with single Gamma pre-coding correction method, the proposed method can further improve phase error compensation about 30%, and the measurement accuracy increase by about 15% in the 3D data.

A three-dimensional (3D) shape measuring system has been built. In the system, a high-speed digital projection module was developed with a DLP4500 chip and synchronously trigged with a CCD camera. Three different frequencies fringe patterns were projected onto the surface of a tested object in turns and were used to calculate the natural phase distribution, as well as the height distribution of the measured object by phase-shifting algorithm and heterodyne phase unwrapping method. Several practical experiments were carried out some tested samples in a small field of view (8 mm×6 mm). The results show that the system can restore the objects’ (coins for example) 3D shape (~100 μm) with a high accuracy.

For researching the pattern of skylight polarization, a real-time measurement system which can capture images of different polarization direction at the same time was designed based on the analysis of existing measurement systems. Based on system light path simulation, the equipment was built and optimized. The error analysis and calibration was made for the equipment and the measuring accuracy was improved. The error of linear polarization degree is less than 1.6%, and error of polarization angle is less than ±0.32°. Outdoor experiments were carried out after calibration. In different bands, the results between the pattern of skylight polarization and Rayleigh model are significantly different. In 520 nm band, the data is close to Rayleigh model’s result.

As a result of the thermal deformation and mechanical vibration on satellite structure, the payload’s pointing regarding to the precision standard will be rotated with a small angle. In this background, a measurement method based on laser Position Sensitive Detector (PSD) technology for this purpose was suggested and a prototype was developed. This method uses unparallel double-beams, double light sources, double mirrors, and double sensors. And the measuring device has compact structure, small volume, light weight, and flexibility for layout. This experiment result shows that the measuring system is very suitable for satellite application and can meet the measurement requirement.

In the testing of aspheric surface, the method of spherical compensation testing is always applied widely. Referring to the limitation of this method, the conditions are put forward which the aspheric surface has to meet. According to aberration theory and Rayleigh criterion, we could get the principles of applying the spherical compensation testing method to test convex quadric surface. And according to the compensation tests with the beam incidence at a distance, we can improve the method of spherical compensation testing. After accomplishing the manufacture of convex quadric surface mirror, the finial consequence shows that the theory and practices in this article can determine that spherical compensation testing can not be used to the test of this convex hyperboloidal surface mirror and δPV of the surface is 0.158 9λ which is better than λ/6 (λ=632.8 nm) under the use of combined compensation testing.

The laser four-quadrant location system has the advantage of small-size and high-precision. Aiming at the disadvantages of structure-complicated, large weight and high-cost of traditional laser four quadrant location optical system, the Fresnel lens is used to take the place of the ordinary lens. Based on the principle of laser four-quadrant location optical system, effect of spot illumination distribution on ratio between four-quadrant’s deviation signal and the azimuth angle is analyzed, conclusion that only the spot whose centre illumination is lower than edge illumination could meet the ideal liner relation is obtained. The design method of ideal laser four-quadrant location optical system is given, a four-quadrant location optical system which can testify the analysis result is designed using this method. The ratio between deviation signal and azimuth angle of simulation result could approach a constant value.

A 5 mega-pixel lens system is designed with ZEMAX. The mobile lens is composed of 3 plastic aspheric lenses. The F-number is 2.8 and FOV is 64°.The sensor OV5645 has 5 mega-pixel which is made by Omnivision, whose pixel size is 1.4 μm×1.4 μm, so the limiting resolution is 358 lp/mm. The final design result shows the optical total length is 4.163 mm, the MTF value is more than 0.3 in full fields of view at 1/2 limiting resolution, the relative illumination is more than 55%, the distortion is less than 1%, and the field curvature is less than 0.1 mm. It possesses an excellent imaging performance.

A coded exposure camera based on the technology of hardware camera is proposed to get the unambiguous image of fast moving objects. A binary code sequence is pre-set in the sequential circuits of CCD to control the pattern of exposure and the imaging sensor in CCD is driven by FPGA. The deconvolution algorithm is applied in a microprocessor of ARM to deblur the image in the next step. The high frequency information could be preserved by using the code sequence to control the exposure pattern. Therefore, a high quality image could be obtained thanks to the fact that the deconvolution algorithm of deblurring could be converted from an ill-posed problem to a well-posed one. The experiments show that the images taken by the coded exposure camera are clearer and better than the images taken by traditional camera.

The simulation of the light-guide structure of the edge-lit LED backlight source and the influence of the practical factors on the light source are studied. Firstly, the main functions and characteristics of the edge-lit backlight light structure are briefly analyzed, and the corresponding theoretical basis is provided. Then, a simplified and accurate light guide structure simulation model is proposed based on the optical characteristics of light-guide structure, and the model is simulated with the optical software of TracePro. Finally, an in-depth discussion and analysis on the characteristics and influences of light-guide structure of the LED backlight source is presented with the actual processing practice. The simulation results are verified by corresponding experiments. The influence of the essential factors on the effect of the backlight source, including the matching between the light guide structure and the LED light source, making the light guide plate thinner and designing dot scheme for high efficiency and good uniformity, is emphatically analyzed for LED backlight system. This research may offer some practical significance, for the reason that it proposes a reasonable and optimized design for light guide parameters and alleviates potential mistakes in the process of actual production.

To ensure a proper temperature rise of a Digital Micro-mirror Device (DMD), the thermal design of DMD was studied by calculation and experiment. A forced liquid-cooled thermal design was presented, and the temperature rise of each measurement point on DMD was calculated. Calculation results show that the maximum temperature rise of the DMD reflecting red light is 14.4℃, the maximum temperature rise of the DMD reflecting green light is 11.3 ℃, and the maximum temperature rise of the DMD reflecting blue light is 10.2 ℃, which have met the working temperature requirements. Thermal design and calculation are also verified through a thermal test and the maximum deviation between thermal calculation and thermal test are not exceeding 2.5 ℃. The thermal tests verify the reliability of thermal design and the correctness of thermal calculation, and supply a theoretical warrant for thermal design optimizing of DLP digital cinema projector.

In the experiment, laser cutting was performed on stainless steel and carbon steel by using high brightness diode laser as light source. The experiment verified the feasibility of the diode laser cutting, and analyzed the cutting process comprehensively. When using oxygen as auxiliary gas, most cutting energy was from the oxidation, which was promoted by laser. When using nitrogen as auxiliary gas, the main factors that influenced the cutting results included the total output energy and the laser power density. 500 W laser output power was suitable for sheet metal around 3 mm, and speed of diode laser cutting was found similar to the fiber laser in addition to the similar cutting surface and slit shape. The cutting ability of diode laser can be further improved with an optimization scheme of oxygen aided cutting technology for carbon steel, which can success in cutting 6mm thickness carbon steel.