The design of infrared optical window for the ultrasonic pod infrared system loaded in the plane is introduced.The principles of Focal Plane Array(FPA) and infrared imaging are included. The infrared images of object in flight are produced. A great deal of temperature data gathered from tracking target are offered and analyzed. A new measurement to the characteristic of the object flying in the air is achieved from the study of the results.

The improvement on Model HWRX-3 infrared camera is presented, which includes the design and development of a real-time digital frame grabber and the hardware driver program under Windows 98 operating system. A powerful software package with many image processing functions is also successfully developed. It is effectively applied to ultrasonic infrared thermography (UIT) technique, which is considered as a new NDE technique. Moreover, a useful infrared thermography image procession method is put forward for UIT.

The relation model between the energy exchange of buried mine and variables such as solar irradiance, atmospheric radiation flux density, self-emitted radiation, heat convection and evaporation in background environment is discussed. A set of practical survey data is analyzed. Finally, the explanation of the quantitative analysis using the model is given.

A novel uncooled silicon capacitive infrared detector with high speed and sensitivity is introduced.The design idea and fabrication processes of this infrared detector are described in detail. The detector operates on infrared absorption by special gas sealed in a micro-cavity, which results in its micro-capacitance change. All fabrication processes are realized by MEMS technology. The experimental results indicate that the prototype samples have good sense of infrared signal.

The method to diagnosis inner wall defects by combining infrared technique and inverse heat conduction calculation with computerized axial technique is a challenging subject because of difficulty in mathematical theory and being confined by geometrical shape. The inner wall defects at corner of a 3D spheroid are located using this method in this paper. The general analytic equation of inner surface is given based on the exposed surfaces temperature data obtained by scanning, which is as an extra bounder condition in the inverse solution to find inner boundary geometry. The size and position of defects can be determined from that. The feasibility of the method is also proved. It is a diagnosing foundation for this kind of equipment and it can be applied to other situations when the second or the third conditions are prescribed along the inner boundary.

A novel configuration of tunable single-longitudinal and single-polarization erbium-doped fiber laser is presented, while two stages of quasi-collinear acoustooptical tunable filters(AOTFs) are used as the tuning elements. An unpumped Er3+-doped fiber(low dopant) is used to narrow the linewidth and eliminate mode hopping. The theoretical calculation shows the output power is 5.5 mW and the linewidth can reach 10-4 nm level when the pump power is at 100 mW. All these data are significant for the oncoming experimental study.

According to the technical characteristic of pulse matriculate wave-gate in the seeker of laser semi-active guidance weapon, a fooling interference program is put forward which applies high-repetition-rate DPSSL to interfere the guidance system. The technical principle of the program is analyzed, and the major qualification is argued and analyzed preliminarily.

There are essential differences in the fractal characteristic parameters between the natural background and man-made targets, which provide a new kind of thought for target detection. The fractal application for the target detection is discussed. The fractal application always bases on the various fractal parameters, which sometime relies on fractal parameters completely, sometime combines the fractal parameters and other parameters each other, and sometime combines the fractal method and other methods each other. It is studied on the estimating methods of some fractal characteristic parameters and their application in the target detection.

According to the features of infrared point target in the sky, a detection algorithm is presented. Firstly, contrast of the images are enhanced by high-pass filter, and then the coarse detection is performed based on differentiating two neighboring frames in image sequence to extract candidate target regions. Secondly, the original image or the enhanced image is segmented in the candidate regions, so the possible point targets are detected. Finally, the true target can be accurately detected by removing noise interference based on continuity of moving target in the image sequence. The effectiveness of the algorithm is proved by the experimental results for several real infrared image sequences.

Detecting targets from infrared cloud background is a typical application of infrared imaging techniques. In order to detect small targets from infrared cloud background, the background properties were analyzed, the infrared background model was built, and an adaptive background suppression algorithm based on Wiener filter was put forword. After getting rid of the fake targets and targets segmentation, the targets were detected and tracked. Experiments were done in infrared cloud background containing simulated targets. The experimental results show that this algorithm can effectively detect targets from infrared cloud background image sequences whose SNR equal to 2.0.

A new method for automatic vanishing point detection on the image plane is proposed, which is based on a combination of line clustering and cross point statistical testing. The knowledge of vanishing point is an important step for oncoming camera self-calibration and 3D reconstruction. Under perspective projection, lines paralleling each other in space will converge to a point called the vanishing point in the image plane. Straight lines are automatically extracted using Burns line finding algorithm, and line clustering is applied to detect groups of lines that intersect on the same vanishing point. The vanishing points are automatically detected and it doesn't need priori information on the geometry of the related lines in object space. The precision of the end point coordinates is set to sub-pixel. The method is applied to the real images and the experimental results illustrate that the algorithm has successful performance and robustness against noise.

A design of Gabor filters which are distributed as a ring on the frequency domain is proposed. In the sense of energy, the selection procedure of adaptive filtering ring and feature extraction expression after Gabor filtering are presented. After multi-orientation filtering with the Gabor ring, texture features of every pixel are derived and segmentation is performed. Experiments show that better results can be obtained.

Based on statistical feature of image blocks, a method for infrared object extraction is proposed. Firstly, the entire image is divided into blocks with scale decreasing gradually. According to the maximal value of the functions composed of statistical feature, seed region of the object and confined region that contains the entire object are obtained. Within the confined region, a fast method of region growing is applied to the seed region, and infrared object is extracted finally. Infrared aircraft images of varying object size are experimented to demonstrate the effectiveness of the proposed algorithm. Consuming time is also estimated if the proposed algorithm is implemented on the digital signal processor TMS320VC33-150, and it meets the requirement for real-time object extraction

Image registration is an important subject in computer vision. How to improve the probability of registration and precision of localization is a hard task. A product correlation algorithm of image registration based on feature and spatial-temporal relationship is put forward. Firstly, extracting edge features of image using wavelet for sequential real-time images and base image. Secondly, the sequential real-time images (3 frame) are matched with base image using normalized cross-correlation, and the registration points searched in sequential correlation matrix are fused according to the offsets. This method overcomes the shortcoming that the probability of registration is decreased with area increasing because of geometric distortion, and improves the robustness of registration.

The image singular value vector has invariance about the transformation of translation, rotation, etc. It is considered as a kind of image algebraic character, applied in image matching and image recognition widely. By matrix operation, the invariance of image singular value vector about the reducing and enlarging transformation is proved, and this character is applied to the image matching. The experimental results show this invariance is an important character of the image singular value vector as same as other invariance about the translation and rotation transformation.

The traditional method via statistical characteristics is usually used to describe terrain features. However, it will fail to discriminate some regions with similar statistics. A fractal method based on texture and structure of terrain is presented. The object described by this method is multi-scale structural feature. According to the characteristic of human vision and starting with inherent mechanism of terrain, the method embodies the procedure of gradual analysis under multi-resolutions

A new active mesh generation method based on image content and mesh node motion vector estimation approach are presented. The mesh-design procedure ensures that the edge of active mesh conforms to objects' edges, which will avoid the failure of motion estimation if different moving objects exist in the same triangular patch. Active mesh can ideally overcome the block artifacts, because it is a continuity-preserving transforms. Active mesh can get more reliable approximation of the real motion field even on local moving details.

Some subjects of image fusion based on wavelet transform are discussed. Firstly, the method of selecting wavelet bases in image fusion is proposed, and the principle and formula on phase adjustment are deduced. Secondly, a new fusion rule of high frequency channel is presented. Based on the rule, a series of experiments are made on PC. The experimental result indicates that the algorithm and fusion rule are very suitable for multi-resource image fusion, especially for multi-spectral image fusion.

The purpose of image fusion is to merge complementary information from multi-sensor data for visual perception, such as in surveillance and reconnaissance. According to the theory of multi-resolution analysis of wavelet transform, a new approach of image fusion is proposed. Experimental results show that the image fusion method based on multi-resolution analysis is very effective and can be applied to wide research fields.

The main problem of vibration blurred image restoration is how the restoration can be done under loose constraints. An effective and fast algorithm is proposed after algorithms of this field are investigated. Pseudo whitening method is applied to obtain Modulation Transfer Function(MTF) corresponding to Point Spread Function(PSF). Then PSF is obtained by solving equation set based on minimum phase theory. Under few constraints, the algorithm avoids the complexity and time-consuming of traditional iteration method. Experiments show this algorithm has high the efficiency and the ability of anti-noise.

Combining image neighboring relativity with fuzzy theory, a new fuzzy mapping technology is defined. Based on this technology, dynamic adaptive enhancement is made towards the high frequencies and low frequencies of the mapped image. Experimental results illustrate the effectiveness of removing noise and preserving details of the target

The shortages of current three methods of residual nonuniformity (residual error) calculation for infrared focal plane array are analyzed in aspects of theoretical premises, calculation methods and results, etc. Based on the response nonlinearity of infrared detector, a new algorithm is proposed. It is proved that this algorithm is more concise and more accordant to experiments. In comparison with the previous theories, a proximate formula is also deduced.

Within the scope of visible spectrum, the color of high-temperature object changes with the temperature. A method of temperature measurement using image color based on neural network is proposed. Hue and saturation of the HSV model are used as the pattern character vectors. BP neural network is used to approximate the nonlinear relationship between the color and the temperature of the high-temperature object. The experimental result shows that the accuracy is within acceptable scope, and the result can't be effected by the environment, so the method is feasible.

In the existing face recognition methods, recognition time usually increases linearly with the scale of face databases, which will certainly make the recognition algorithms less effective in practice. A novel face recognition scheme is presented. Firstly, a group of typical faces is selected, called reference faces, as comparison foundation through clustering in the eigenspace. Then, a similarity sorting space is built up based on the similarities with reference faces. For recognition, only need to compute the similarity of the input face with the reference faces to pick up the most similar faces in the sorting space quickly, then to compute the similarity with these picked up faces to determine the final recognition result. Since a lot of calculation is done offline, the recognition speed is improved greatly. Through selecting the number of reference faces and the mostly expected similar face set, the recognition time for different scales of databases can be controlled approximately. Experiments show that the recognition capability of out scheme decreases a little than ordinary methods,but the recognition speed is increased greatly.

The principle of determining 3D coordinates of a point in the space using multi-joints 3D scanner is described. The error metric between the real value and the designed value of structural parameters of multi-joints 3D scanner is analyzed. It is pointed out that the calibration problem is measuring error minimization problem over all possible values of the parameters, and a genetics-based calibration algorithm for the structural parameters of multi-joints 3D scanner is presented. The experiments show that the proposed method is effective to calibration.

Multi-grid method is used to solve the 3D electrostatic field distribution problem. The results to calculate concentric spherical system and len of bi-cylinder-cathode indicate that computational accuracy is 10-3～10-4. Comparing the two results in view of computational efficiency and computational accuracy , it appears that MGM is superior to FDM in solving electrostatic field distribution for the electron optics problem and it can greatly improve the computational efficiency of field distribution. A firm ground is provided for the design of non-rotation symmetry tube.

A high-throughput popeline variable length coding strategy is presented, which encodes run-level coders individually by two steps look up tables and combines them to form steam. The presented strategy uses lesser memory space and has favourable expansibility by two RAM look up tables. The result of FPGA simulation shows that the presented strategy adapts to high-throughput data HDTV encoding.

A novel method of logo insertion in MPEG-2 video is proposed. Compared to the method of Decompression-Addition-Recompression, it needs neither DCT and inverse DCT transformation nor motion estimation. Therefore, computational complexity is reduced and it is quite suitable for real-time procession. The experimental result shows that the proposed method has almost the same effect as pixel domain processing method.