Research On Satellite Target Identification And Feature Extraction Method

The use of space-based detection system to detect spatial target has been attracting much attention. Its application model determines the location and weather conditions are not restricted and it has the good effect of the detection, the strong ability of the surviving. But because of the finite ability of the signal processing in the star, the power can't compare with the ground monitoring system, and in the most time, it has the lower signal to noise ratio. Then the huge effect is generated by the noise and the other interference factors. It is difficult to distinguish decoys and target, when decoys have the same orbit and similar shape with target. The radiation difference between the target and the decoy had been proved in researches. The purpose of this paper is how to extract the radiation difference between target and many kinds of interferences, and how to identify the target. By analyzing the advantage and the disadvantage of many kinds of existing identification methods, the identification method is that the radiation difference is the main, and track difference is a supplement to the spatial target identification. The main content is including the following:(1) Radiation composition of the spatial target and its difference from interferencesSatellites were studied through establishing the relative positions model, and then spatial target radiation calculation method was determined from the principles of radiation. The radiation composition of the spatial target was determined, and the influences of many parameters such as different detection angels, detection distance were analyzed on the spatial target radiation. Next was to calculate the stars radiation in detection bands, and select stars influencing the target identification by deriving the stars radiation function. The main components and distribution of noise were analyzed to determine the influence on the target identification. Based on the existing radiation difference between target and decoys, the differences between target and stars, the system noise and other interferences were determined.(2) Under noise interference, target radiation extraction based on the subband equivalent blackbody temperature (SEBT) First the concept of SEBT was proposed, and then the SEBT of spatial target was obtained by calculating the radiant exitance ratio and the R(T) curve of targets with different temperature in different bands. The use of the SEBT to extract the target radiation wouldn't be affected by the detection distance and detection angle, and could effectively reduce the interference of the space environment and fully showed the radiation. Under noises interference, the calculation of the radiant exitance ratio had error, and the observational data were used to design the effective estimator to minimize the impact of noise. Through the analysis of the relationship curve between estimation error and the radiant exitance ratio, the estimator was further amended to improve the estimation accuracy. Finally, multi-frame estimation method was analyzed, which could further improve the estimation stability of the SEBT.(3) Determination of spatial target's optimal detection bandsExtracting accuracy of radiation characteristics is relative with noise level and detection bands choosing. By analyzing the trend of SEBT calculation error under noise interference, the best detection dual-waveband was determined, and the higher calculation accuracy was got. In order to reduce the calculation error further more, calculation error expression of different detection multi-bands combination was analyzed to get the principle of error further reduced. Finally, the effect of the number of detection bands combinations was analyzed, and the conclusion was not the more detection bands the better.(4) Spatial target identification based on the extraction of radiation differenceIn order to remove the effect of noise, dual-threshold segment were used to deal with IR images based on radiation intension and radiation relation. Further, after the SEBT sequences of the target were calculated, the variation frequency and distribution variance of SEBT would be obtained by wavelet analysis. The trajectory tracking method assisted these two characteristic parameters in target identification. Simulation experiments validated the method of target identification and show its value, especially when decoys had the similar shape and the same orbit to the spatial target.(5) Imaging satellites feature extractionWhen the target became constantly near the detector, the satellite imaged area target. The need for the extraction of characteristic parameters is to determine the position of interesting region. After contour extraction and similarity calculation, observation direction to the target was determined. The improved algorithm was used to extract corners of satellite images, which determined the interesting region according to the position of every contour.In this paper, these researches lay the necessary groundwork for spatial target detection and identification by space-based detection system, and play a role in space detection engineering in our country.