Research On Sea Surface Infrared And Visible Image Registration Technology

As China's national strength grows stronger,maritime activities become more frequent,and maritime safety accidents also occur more frequently.Compared with other safety accidents,marine safety accidents are very sudden,and most of them occur in severe weather such as heavy winds or fog.Therefore,when searching and rescuing people in distress at sea,the infrared imaging system is the most suitable system for searching for sea targets.Cooperating with the visible light imaging system to complement the image information of the scene of the sea distress target,it can improve the detection capability of the sea distress target and improve the effectiveness of the search.Because there are few feature points in the sea image,it is difficult to realize the high-precision registration of the sea infrared image and the light image directly using the existing registration algorithm based on the image features.Therefore,this paper proposes a registration method based on the photoelectric stable platform.Registration using the characteristics of the photoelectric stabilization platform is mainly divided into two steps:The first step is to calibrate the relative positional relationship between the infrared and visible light cameras installed on the photoelectric stabilization platform.Determine the registration parameter curve of the two cameras when imaging the target scene,and store the curve in the system database.The second step is to read the registration parameters from the system database to register the infrared and visible light images according to the shooting distance of the sea surface image when searching for the sea surface target.First of all,through the analysis of the characteristics of the photoelectric stability platform,the design,installation and use requirements of photoelectric stabilization platform are put forward.In order to ensure the detection resolution of the sea surface target,the matching relationship between the shooting distance,focal length and scanning field of view of the infrared and visible cameras is analyzed,and the registration parameters of the images taken by the two cameras in the ideal state are simulated,so as to verify the reliability of the registration parameters of the images taken by the two phase cameras in actual use.Then,according to the use requirements of the photoelectric stable platform,under the condition of different shooting distance,the infrared and visible images with more feature points of the same target scene on the ground are registered to obtain the position relationship of the two cameras.In order to ensure the high-precision registration of infrared and visible image,this paper proposes an improved SHIFT-FLANN algorithm.On the basis of RANSAC algorithm for feature point selection,the mean iteration is added to make a more detailed selection and decision of matching point pairs,which improves the accuracy of image registration.Finally,based on the improved SHIFT-FLANN algorithm,the infrared and visible images captured by the photoelectric stabilization platform built by the research group are registered to obtain accurate registration parameters.The registration parameter curve is fitted according to the shooting distance and stored in the system database.When searching for the sea surface target,the registration parameters are read from the system database to complete the registration of the infrared and visible light images directly according to the shooting distance of the sea surface image.Experiments show that the method proposed in this paper has high registration accuracy,small calculation amount and fast processing speed,which is very suitable for engineering applications.