Adaptive confocal laser ophthalmoscope with the high-resolution and dynamic imaging ability has been widely applied in specific biomedical and clinical medical fields. In order to apply the noncircular pupil filter and other pupil modulation technology without influence in wavefront detection, the system needs two light sources for imaging and aberration correction respectively. This paper has designed an adaptive optics scanning laser ophthalmoscopy with two sources, and then analyzed the differences of the aberration of the two light sources. Then, the aberration correction and high-resolution imaging ability of the system have been verified，and the brightness, contrast and resolution of the image have been significantly improved after close-loop. Finally, we have studied the feasibility of realizing the dark field imaging by semi-circular pupil and obtained the bright and dark field images of the artificial eye.

The modular design and multi-channel integration has become the main thought of developing the pho-toelectric equipment, and the multi-axis parallelism directly influences the equipment performance. The current me-thods cannot meet the actual testing needs of multi-spectral, multi-axis, high-precise and large axis space. Thus a multi-spectral and multi-axis parallelism testing scheme is put forward by adopting the designing thought of reflective type and optical axis translation. The reflective collimator is designed to solve the multi-spectral and multi-axis par-allelism testing problems, and the optical axis translation design can increase the axis space of multi-axis parallelism test. The results show that the parallelism testing error is less than 0.134 mrad and the axis space can reach 0.5 m, which can satisfy parallelism testing needs of most photoelectric equipment.

Comparedwith mercury-vapor lamp, ultraviolet (UV) LED suffers from the disadvantage of having a singlewavelength, which is not fully compatible with the existing photoinitiator of ultraviolet ink. It is difficult using UVLEDs to achieve the curing effect comparable to what mercury-vapor lamp can do. This article presents a design of UVLED ink curing system that provides evenly mixed light with multiple wavelengths. We put a UVLED array with three wavelengths on a cambered surface and achieved uniform illumination in ink curing area by using optical freeform surface. By adding the adjusting of the tilt angle, we solved the dilemma of uniform wavelength mixing and uniform illumination. The ray tracing simulation results show that an illumination spot with uniform wavelength mixing, an average illumination of 110.7 mW/cm2, and an illuminance uniformity of 0.82 is obtained on a target surface 600 mm away from the light source. This design is expected to simulate the multi-spectral illumination effect of mercury-vapor lamps, which can promote the application of UVLEDs in ink curing.

The 3D point cloud data obtained from the laser line structured light scanner has redundancy,and a point cloud simplification algorithm based on the two order non-uniform partition is designed and implemented to deal with locomotive running department in this paper. First, according to the intrinsic shape signature (ISS), the point cloud normal vector of the detected object are estimated and the feature points of the point cloud are extracted. Then, according to the distribution of the feature point cloud, the point cloud is first divided non-uniformly to obtain uneven initial cloud patches. Finally, the divided cloud points are mapped to different Gaussian spheres for further subdivi-sion. The mean shift clustering is performed on the Gauss sphere to extract the center of gravity of each cluster in the actual three-dimensional space. The set of the center of gravity is the result of simplification. Experimental results verified the effectiveness of the proposed method. It can keep the details information of the point cloud while en-suring a high simplification rate. Comparing with the existing method, this method balances the speed and accuracy, and is more suitable for the on-line locomotive automated detection system.

Aiming at the poor adaptability of traditional lane recognition method in complex pavement, this paper proposes a lane recognition method based on full convolutional neural network and conditional random field, ac-cording to image segmentation technology. The method can make the neural network model identify the lanes by training a large amount of data, and then make the segmentation of the lanes' coverage and the lane edges more perfect through the conditional random field. At the same time, in order to solve the high requirement of real-time detection in expressway, a fully convolution neural network is designed in this paper. The network structure is simple with only 130000 parameters and three improvements are made as follows: BN algorithm is used to improve network generalization ability and convergence rate; LeakyReLU activation function is used to replace the commonly used relu or sigmoid activation function, and using Nadam as the network optimizer makes the network have better ro-bustness; Conditional random field is used as the back-end processing solution insufficient lane segmentation and further to increase lane coverage. Finally, in order to solve the problem of complex road environment in urban road testing, this paper uses the back-end processing of FCN-16s network model and conditional random field to realize the recognition of complex urban roads. Experiments show that the network model designed in this paper is more real-time and sufficient for lane identification in the face of high-speed expressways and simple lanes. In the complex environment of urban road, FCN-16s model plus conditional random field can identify lane more accurately and get good results on KITTI road test benchmarks.

Pressure-sensitive paint technology is a wind tunnel pressure measurement frontier technology with high economical efficiency and high speed. In the wind tunnel test, due to the strong wind, the model will be distorted, making the wind image and the windless image difficult to register, which will seriously affect the pressure mea-surement accuracy. In response to this problem, this paper innovatively applies the two-dimensional non-rigid iterative closest point (ICP) algorithm to solve this problem. The point cloud method is used to make the image detail area to be effectively registered, and it is also conducive to the subsequent three-dimensional reconstruction work. However, due to the two-dimensional non-rigid ICP algorithm, only the two-dimensional coordinate positional rela-tionship is considered. The correlation of the pixel grayscales of the pressure-sensitive paint image is neglected, so that the registration accuracy is not too high. However, if the three-dimensional non-rigid ICP algorithm is directly used, misregistration will occur. Therefore, in order to further improve the registration accuracy, this paper proposes a non-rigid ICP algorithm based on pixel-based search strategy. The algorithm designs a dual-target search strategy that takes 2D coordinates and pixel gray values into consideration and achieves accurate local matching, realizing point search and double goal optimization. The algorithm of this paper is compared with five registration algorithms on multiple sets of pressure sensitive paint images. The experimental results show that the proposed algorithm has the best registration accuracy. Compared to the suboptimal algorithm, the RMSE is improved by more than 15% and the NMI is increased by about 5%.

When measuring the oil concentration lower than 10 mg/L in reinjection water based on ultraviolet fluo-rescence method, high thermal power of the light source will reduce the accuracy of measurement results. To reduce the thermal power of light source, pulse modulation dimming method (duty ratio less than 50%) was proposed to replace square-wave modulation dimming method (duty ratio equal to 50%). Two noise analysis models for pulse modulation and square-wave modulation dimming method were built. Signal-to-noise ratio (SNR) of measurement results for two methods were compared. The relationship between SNR and light source power was derived. The optimum choices of current amplitude and duty ratio for pulse modulation dimming method were proposed under the condition of the same signal-to-noise ratio. Results show that the light source power for pulse modulation dimming method is less than 21% of that for square-wave modulation dimming method under the condition of the same sig-nal-to-noise ratio. Finally, signal demodulation models of two dimming methods were simulated by MATLAB software. The simulated relationship of SNR for two methods is consistent with the theoretical analysis results.

A new type of spectrosensitometer has been developed, which is characterized by a wide spectrum range of 340 nm ~ 900 nm and a large exposed area of 202 mm × 90.5 mm with multi-step light intensities on it. The optical density value error of each step on the 18-step wedge with high precision is not greater than 0.01. The film filter evaporated according to the spectral characteristic of the light source can eliminate the secondary spectrum of grating. The automatic control acquisition system is developed by LabVIEW and all-in-one PLC with HMI. In the ho-rizontal direction, the grating displacement sensor is adopted to form the closed-loop control, and the wavelength positioning deviation is less than 0.05 nm. Linear compensation method is adopted in the vertical direction with a height deviation of less than 0.05 mm. The spectrosensitometer automatically measures the optical power per unit area of lights with different wavelengths and light intensities on the step wedge. The shutter controls exposure time. Photosensitive materials are once exposed within the scope of the wide spectrum. After being developed and fixed, the optical density value can be measured by densitometer. The spectral sensitivity curve of a photosensitive ma-terial with a certain optical density value can be drawn according to the national standard (GB10557-89).

The axis consistency of multiple optical sensors is an important guarantee to ensure the normal operation for photoelectric task equipment of weapon system. The presented status quo of methods and equipment are ana-lyzed for measuring the consistency of large spacing axes. An axis consistency detection method is proposed based on non-cooperative target image processing technology. Specifically, it is available to select scenes with typical characteristics in the far field as non-cooperative targets. Then, the axis consistency detection results are obtained by comparing the position differences of non-cooperative targets in different image spaces. Compared with other detection methods and equipment, our method avoids many disadvantages including huge volume, heavy weight and the limited operation environment. Furthermore, it is especially suitable for axis detection of in field and on-line to large-distance photoelectric equipment, which shows a bright application prospect.

The complex background and laser stripe noise affect laser stripe extraction. Adaptive double threshold segmentation method and the improved gray weight model are proposed in this study. First, the characteristics of the laser stripe and the source of noise in the image are investigated. Bilateral filter is applied to remove the noise of images. Subsequently, the gray histogram of laser image and the double threshold are computed. By sub-regional processing, initial stripe center and stripe width of binary images are obtained. Finally, the sub-pixel center of the laser strip is extracted by the proposed model. The double threshold segmentation method and the improved gray weight model are compared with the traditional algorithms. The results show that the double threshold method is more accuracy in extracting the laser stripe region than the extreme value method and the Otsu method. Comparing with the residual value of sub-pixel center, the improved gray weight model (0.23) has better results than the gray-gravity method (0.71), the extreme value method (0.86), and the Gaussian fitting method (0.86). The algorithms proposed in this study avoid the impacts of complex background and laser stripe noise, increase the accuracy of the laser stripe center positioning and extract the stripe center extraction fast and accurately in complex backgrounds.