Research On The Key Techniques Of Photoelectric Target Detection Under The Complex Environment

The precision and accuracy of photoelectric detection system have been affected seriously since the technology of photoelectric false targets and camouflage is more wildly used which makes the modern battlefield environment increasingly complex.How to distinguish the real and the false targets fast and accurately in the complex battlefield environment is one of the important tasks and urgent demand of the development of modern photoelectric detection system.Based on the infrared radiation optical properties and the polarization reflection optical properties,further researches on the contents of three aspects including nonuniformity correction for detection system,infrared decoy false target discrimination and polarization camouflage target detection have been launched in this dissertation around the technology of photoelectric target detection in the complex environment.The nonuniformity has been the important problem which affects the infrared detector's performance.The multi-channel polarization detection system is also faced with similar problem of channel response's inconsistency.These nonuniformity problems have seriously affected the accuracy of the detection results.Firstly,the generating mechanism of ghost in the traditional constant statistical algorithm has been analyzed in detail for the ghost problem of infrared nonuniformity correction algorithms.A scene-based nonuniformity correction algorithm based on temporal median filter is presented utilizing robustness of median filter for the outliers and the performance of ghost removing has been improved obviously.Then,an adjacent differential statistics nonuniformity correction method is presented.The algorithm move the traditional perspective of intensity signal to the new perspective of the difference signal between adjacent pixels.The nonuniformity compensation is obtained by calculating the statistics in the differential domain.The proposed algorithm makes the least residual ghost in the correction results with excellent convergence speed.Finally,a scene-based polarization constant statistical non-consistency correction algorithm is presented which calculates the channeles' parameters using the scene information from the multi-channel polarization system under random rotation moving.Compared with the existing correction algorithm,the applicable conditions of the proposed algorithm is more common and the calibration results are more accurate.Based on studying different radiation characteristics of real infrared targets and false decoy targets in multiple perspectives of features,several new feature descriptions aimed at distinguishing between target and decoy are constructed.A dynamic contour moment is proposed as a feature description using variation of the target and decoy's contour moment.The contour moment and curvature autocorrelation fractal features are presented with contour fractal characteristics.A special description of motion-radiation ratio is presented using the relationship between radiation variation and motion property of the infrared decoy.The model of dynamic target detection in the complex environment is established based on multi-features joint probabilistic data association according to the multiple features constructed.The model can make the optimal discrimination for target and decoy in the complex environment.Finally,the infrared target detection experiment is carried on in the process of releasing infrared decoys from target.It has been proved that the dynamic target detection model based on multi-features joint probabilistic data association can discriminate the target and decoy accurately and effectively under the countermeasure condition of decoy false target which makes that the system has stronger ability of counter-countermeasures detection in the complex environment.The polarization optical characteristics theory and the polarization distribution model of rough surface have been studied with the reflection polarization properties of different materials of true,false and camouflage target.Then,a target detection method based on multi-angle polarization information is presented.In this method,the polarization data in different angles between the directions of incident light and detecting is used to recover optical constants of target materials.The target detection is launched with the optical constants of different materials.What's more,an adaptive polarization target detection method based on Mueller matrix is proposed.The method regards the polarization states of the detection system as a classification hyper-plane innovatively.The optimal separating hyperplane can be obtained through learning the Mueller matrix information of the target and non-target in the scene.The polarization states can be adjusted to the optimal classification status adaptively so that separability between the target and non-target in the polarization image can be maximized.It provides a new idea for development of the intelligent detection equipment in the battlefield.