A dynamic calibration method was proposed to determine the relative orientation of the axes of the strapdown inertial navigation system (SINS) and the CCD star sensors (CSS) for shipboard SINS/CSS system at the rigid and flexure installation conditions. Specifically, the Kalman filtering method is utilized to estimate the SINS attitude error and the CSS installation error, in which the observation function is founded by using the attitude data separately measured by CSS and SINS, and the state function is constructed by using SINS navigation error functions. To validate the proposed calibration method, the simulation experiments are studied by using the actual navigation data and ship flexure data measured by Yuanwang Ⅲ TT&C ship. Results show that the calibration can achieve a high accuracy when the CSS and the SINS are rigidly installed together, while the calibration error will increase when the CSS and the SINS are separately installed at different locations for ship flexure influence.

Due to the changes of environment and the axes of optoelectronic tracking system during the pointing calibration by stars along with the time, people cannot get a very precise result all the time. The problem is considered and the algorithm based on WLMS is introduced to meet the needs of two types. One is to get precise calibration result in a small area of the sky. And the other is to get the precise calibration in real time guiding. Two simulations are introduced to verify the feasibility and validity of the algorithm by using the data gotten from the field.

Space observation image is easy to be interfered by cosmic rays. A rejection algorithm of cosmic ray in space observation image is presented. Firstly, the stray light is calibrated in observation image by Top-hat transform. After processing, the image background is uniform, and the histogram shape recovered nearly Gaussian. Then, a combination of morphological operations and median filter is used to eliminate the interference from bright saturated stars. Finally, the Laplacians edge detection algorithm is used to detect cosmic noise pixel, and the neighborhood minimum filtering method is used for cosmic ray pixel interpolation. The experimental results show that, the algorithm is superior to other algorithms in two aspects of the detection and false detection probability.

Due to the difficulties caused by large satellite cloud image data with limited transmission channel and storage space, an approach of block compressed sensing of satellite cloud images is proposed based on Tetrolet transform. This approach introduces Tetrolet transform into the sparse representation step of compressed sensing which can depict the detail and texture structure of satellite cloud image well, and combines block compressed sensing with smooth projection Landweber iteration method to accomplish image reconstruction which can improve the computational efficiency. Meanwhile, in order to further improve the quality of reconstructed cloud images, another improvement scheme for the sparse representation of cloud images is proposed. Firstly, a layer of Laplacian pyramid decomposition of the original image is taken, and the low frequency component and high frequency component obtained are divided into blocks and sampled respectively. Then, the low frequency component is represented by Wavelet transform, while the high frequency component is represented by Tetrolet transform, which can not only play the advantage of different sparse representation, but also make full use of the advantages of Tetrolet transform in expressing the important information of cloud images, such as directional texture and edge information. The experimental results show that the reconstruction quality of the proposed method is obviously superior to Tetrolet, DWT, Contourlet and DCT sparse representation methods under the same sampling rate.

Aiming at the problem of water quality anomaly monitoring, a bio-monitoring method based on computer vision and support vector machine is proposed. First, fish behavior movement information is collected by computer vision. Then, establishing training sample set is used for obtaining water quality anomaly monitoring model. Finally, the model is utilized to analyze the fish data of unknown water quality. Kernel function type and parameter optimization have a significant impact on the model. The different types of kernel function experimental results are compared to choose the best kernel, and then Particle Swarm Optimization (PSO) algorithm and Genetic Algorithm (GA) and the Grid Search method (Grid Search) are used to optimize parameter. The experimental results show that the method can monitor the water quality quickly and efficiently.

Aiming at the positioning platform of heterodyne interferometer, a new position method of phase locked has been introduced to simplify the system structure. The positioning system working principle and process have been explained and the feasibility of the phase locked method has been proved. The core link of positioning system, such as high-frequency, phase-shifting circuit and phase demodulation unit, have been discussed in detail. The test experiment of phase demodulation circuit and the whole positioning platform have been performed. The results show that the precision of phase demodulation circuit is 0.07o and the positioning uncertainty is less than 10.1 nm.

In assembly process of cylinder, some factors will cause deformation of cylinder. It affects the match of cylinder and piston ring, and reduces the performance of diesel. In this article, we designed a cylinder-shaped aperture dimensions measuring system based on laser displacement sensor. The dimensions of cylinder both in condition of freedom and preload were measured. Compared with measurement results by traditional measurement tools, it shows good consistency.

In order to obtain three-dimensional data of engine cylinder head blank, a passive line structured light canning measurement system is designed. Calibration of the direction cosine of the movement of camera optical center is given based on principle of virtual binocular stereo vision. Stereo vision system consists of a single camera moving along the one-dimensional rail, which gives Rotation matrix R and Translation vector T between local coordinate system and global coordinate system. And then the origin of local coordinate system is conversion to the global coordinate system so that the line of movement of camera along the rail is fitted and its direction cosine is calculated. Three-dimensional topography of a certain cylinder head is scanned and the cloud data on this topography is obtained, from which the data of Intake port center coordinates is extracted. Finally, the three-dimensional data is dealt with reverse engineering software and surface reconstruction works well.

The mathematical model of structure light measurement system is established, and the factors influencing the measurement accuracy are analyzed. At the same time, a fast new light strip center edge extraction algorithm is used in the image processing method, which improves the extraction accuracy distinctly. The accuracy testing experiments are designed and analyzed by standard plate. The experimental results show that the measurement accuracy of this method is better than that of 0.2mm and 0.3mm, and reached in the detection depth and width on the regular groove, which can fully meet the requirements of sub-millimeter small deformation detection accuracy. It is an effective detection method.

Multi-view 3D data registration is always a research challenges in vision measurement, and a multi-view of 3D data registration method based on global control codeless points is proposed. The global coordinate system is established with many fixed codeless marked points as global control points. The stereo vision combined with structured light measurement system is moved to different locations to measure the object surface of each region, and obtain the local coordinates of 3D data and control points. Automatically obtaining the pairs same named control points of the local and global based on the characteristics of spatial geometry transform invariance of these control points. According to the pairs of control points, SVD algorithm is adopted to solve the transformation relation from local coordinate system to global, and then the global registration is complete. Finally, with 10 circle marked points for the global control points, a shoe last is measured from five different views. The experimental results show that the method can eliminate the cumulative error in the process of multi-view 3D data registration. The average error of coincidence degree is 0.013 mm, the comparison of standard deviation with hand-held scanner is 0.177 mm, and the good registration result can be obtained..

Currently 3D model animation has been widely used in film, television, games and other fields. The basic principle o 3D model animation is that the position of each vertex of the model will change in different time. It can be divided into Morph animation, Skeleton animation and Skinned Mesh animation. Skinned Mesh animation solves the problem of seams. And it is popular because of its realistic, flexible interaction effects. In the application background of three-dimensional e-learning platform, this article describes the principle of skinned mesh animation technology, and compares three kinds of key frame interpolation methods. We hope to improve the picture quality while increasing the run speed of Skinned Mesh animation. Meanwhile, the visual effects of the network teaching platform can be optimized.

Because panoramic video has a 360 degree viewing angle, it is quite suitable for the multi-object detection in intelligent video surveillance systems. First, this paper introduces the DSP-based system hardware architecture. Second, the panoramic image unrolling algorithm and the MHOEI multi-object detection algorithm are given. Then the look-up method of block table on DSP platform for panoramic image unrolling and the resolution reduction method for MHOEI are designed respectively. Finally, the optimized scheme is implemented on DSP processor. Experimental results show that it achieves 120 f/s processing speed for unrolling the panoramic video resolution from 720×576 into 896×180 on the selected DM6437 processor. And after combined with the multi-object detection, the average speed is up to 35 f/s, which meets real-time requirements. Meanwhile, it can detect moving and motionless objects in panoramic video effectively.

Feature matching accuracy affects the precision of image registration, which is one of difficult key points for image registration based on features. In order to solve feature points one-to-many mismatching problem caused by the ratio of the closest neighbor and second closest neighbor from one direction, a bidirectional matching method of features for image registration of visible and infrared image is put forward. Firstly, for enhancing the similarity of two images, image reverse and histogram equalization are adopted to process infrared image, so that more consistent features of high repetition rate are extracted. Next, SURF features are matched bilaterally by using the ratio of the closest neighbor and second closest neighbor, to ensure the consistency between feature matching and reduce matching error rate, and then RANSAC is applied to match feature again. Through the two matches, it can realize precise features matching. The experiment results show that the proposed method is better than traditional SURF feature unilateral matching algorithm based on the ratio of the closest neighbor and second closest neighbor in the correct matching ratio and registration accuracy, and the validity of the method suggested is proved.

Compound array in the DRA area of insects can receive the polarized light in the sky and abstract its polarization information, which enable the insects to navigate. According to the structure of DRA compound array elicited from biological anatomy, we put forward a kind of simplified array architecture made by MEMS technology, which has small volume, light weight, large visual angle and can perfectly obtain the polarization information of the sky all through 180° by combination. Then, we build the model of DRA compound array in ZEMAX, and educe that the material and structure of the model of DRA compound array are reasonable and can be used to detect the polarization information of the sky light effectively.

A simple and practical three-dimensional orientation system is presented based on image formation of planar artificial compound eye. The principle of high-precision orientation with artificial compound eye was given and a design method for the three-dimensional orientation system was proposed. Planar camera array was taken as the structure of the imaging system. Sub-cameras ranged to be a square as sub-eyes of the compound eye. Simple and stable optical geometry were produced by camera array and parallel optical axis arranging, which make the results more accurate. Geometric constraints were brought by multiple directional parallaxes. The calculating model was built by multi-directional and selective stereo matching arithmetic to realize high-precision orientation. A nine-camera sample system was fabricated for testing experiments. The positioning accuracy of the experiment is 2.53e-4 rad.