Based On The Detection And Recognition Research And Implementation Of 3 D Object Structure

Due to real-time performance constraints, corners are commonly cut. Today, however, with steadily increasing computer power, the opportunity exists to explore the full potential of ATR algorithms.Taking advantage of sensors of given spectrum, Auto Targets Recognition(ATR) can capture the image sequence of the target and doing something with the military targets including detection, classification and recognition, which enable the weapon system with the ability of accurate real-time attacks, aiming at improving shoot accuracy and destruction ability of the weapon system. The main working theory of the ATR is as follows:first translate the radiation of the target and background into electricity signal, form and display the target image; then accomplish the target capturing and tracking by doing the real-time matching and recognition with the target image structure, which is prepared before or acquired manually in the loop; at last guide the missile shoot the target.The imaging missile-guiding system requires the ability of real-time and high accuracy of the target detection and recognition algorithm, that is accomplish the task including capture preparation detection and recognition of the target image to reach at the goal of real-time tracking. Meanwhile the preparation of the target image is one of the key techniques, which has a direct relationship with the detection probability and recognition accuracy, it is also one of core components of ATR system.In this paper, the basic experimental data of the research of the target research and recognition algorithm is three dimensional target image. And we constantly improve the adaptability and robustness of the detection and recognition by conducting detection and recognition experiments with the real-time collected images.The first part of this paper introduced the components of the target detection and recognition system and gives a brief introduction of the detection and recognition algorithm, which is thought to be the key technique. This part gave of detailed expatiation of several popular algorithms and concluded the advantages and disadvantages of each algorithm, thus explained the advantages of choosing the algorithm of target structure from the point of view of constraint conditions.The second part of this paper introduced the mathematic and geometrical transformation of three dimensional emulation system in detail. This part centered on mathematic deduction of the definition and transformation of several coordinate system in the procedure of the creation of three dimensional structure images, which includes perspective projection transformation, matrix operation latitude coordinate and gauss ellipse sphere coordinate.The third part of this paper made a introduction of the three dimensional scene emulation. Firstly it introduced the format and regulation of the LINDAR system three dimensional point cloud data source(DSM/DEM), which is required by three dimensional reconstruction technique. Then it made a introduction to the three dimensional scene emulation procedure and introduced several key technique points of the three dimensional scene emulation system particularly. At last, it gave a brief introduction of the general three dimensional software development environment.The fourth part of this paper introduced the image matching techniques based on the matching of the heterogenous images. This section based at the infrared image technique, gave a detailed introduction of theory of several image pretreatment and assessments. And then introduced the application of the three dimensional structure in the domain of ATR and verified the validity of the target structure image by conducting emulation experiments with the real-time infrared image.