Ground Target Recognition Algorithm With Smoke Screen Jamming

Since the 1950 s,electro-optical imaging terminal guidance weapons have been widely used in the battlefield due to its high efficiency and cost ratio.In order to protect the target effectively,smoke screen jamming technology is widely used as an important photoelectric countermeasure.Smoke screen jamming seriously affects the performance of infrared image target recognition algorithm and the strike effect of infrared terminal guidance weapon.Compared with air target and sea target,the background of ground target is more complex and it is more difficult to resist the smoke screen jamming.How to effectively improve the performance of ground target recognition algorithm under the condition of smoke interference is an urgent problem to be solved.Aiming at this problem,the paper has carried on the thorough research in the infrared smoke interference image characteristic,the smoke recognition and the region extraction,the anti-interference strategy design and so on,obtained the better experimental result.The main work of this paper is as follows:In order to quantitatively analyze the influence of smoke interference on the target recognition algorithm,simulation experiments were carried out with smoke transmittance and proportion of occlusion area as variables.Under the guidance of the experiments results,the real infrared smoke interference image sequences of multiple bands were collected through the field experiment.In order to obtain the effective features which can be used to distinguish smoke screen from other ground objects,the spectral characteristics,texture characteristics and dynamic characteristics of smoke screen jamming images are analyzed.A smoke detection method based on motion detection is studied according to the results of smoke dynamic analysis.Aiming at the problem of missing detection caused by motion detection method,a smoke detection method based on super-pixel segmentation and texture feature fusion is proposed.In order to solve the problem that the conventional smoke detection algorithms cannot distinguish thin smoke from thick smoke effectively,a smoke detection method based on smoke transmittance estimation is proposed.This method uses 3D modeling software to generate the smoke image with its own transmittance labeling,which solves the difficulty of calibrating the smoke thickness information in the actual image.By referring to the overall framework of deep learning semantic segmentation network U-Net,the model of transmittance estimation is built.The experimental results show that effective smoke screen regions can be obtained more accurately based on the estimation results of smoke transmittance.In order to use the prior information of smoke screen to counter the smoke screen jamming,a method of anti-jamming strategy selection for target recognition based on the prior information of smoke interference is proposed.Firstly,the smoke screen area proportion is calculated according to the area of smoke screen regions,and then the antiinterference strategy is selected according to the area proportion.When the proportion of smoke screen area is small,the block strategy and position constraint are adopted.The template expansion strategy is adopted for the larger proportion of smoke screen area.In addition,the smoke detection and target recognition are connected in parallel,which is compared with the aforementioned serial connection scheme.The experimental results show that the anti-jamming target recognition algorithm based on the prior of smoke screen jamming can effectively improve the target recognition accuracy with smoke screen jamming.