Multi-source Information Co-processing And Fusion Method

Multi-platform multi-sensor information fusion is widely used in military andcivilian high-tech system. But the observational data of time and space from each sensoroften is not matched in the process of integration. Without time-space registration, therewill be large error in the process of data integration. The the time registration algorithmand space registration algorithm of multi-sensor information fusion is researched in thispaper, and the image fusion method is coverd, too.Firstly, the time registration model, target motion model, registration frequencyselection and time registration algorithm are introduced in this paper. The existingregistration algorithm based on interpolation-extrapolation method and least squaremethod is detailedly derivated and simulated. Then a registration algorithm combiningKalman filter and interpolation-extrapolation method is proposed based on this. Anotherone combining extendedb Kalman filter and interpolation-extrapolation method fordiscrete system is proposed, too. The effectiveness of both algorithms’ is tested.Secondly, the space registration algorithm of offline and online space is researched.The offline registration algorithm is derived by generalized least squares mothod andmaximum likelihood function mothod based on the geocentric coordinate system, andthe registration performance is validated. The online registration algorithm is mainlybased on the geocentric coordinate system of sequential Kalman filter a lignmentalgorithm. Calculating the Jacobian matrix can be avoided by using sequential Kalmanfilter, so the registration accuracy is meted and real-time is improved at the same time.Finally, multi-scale, multi-resolution image fusion method is studied with thecharacteristic of multi-spectral, multi-tempn of the Laplacian oral, multi-sensor remotesensing image fusion process. Through the fusion of multiple sets of images, the futioneffect of Laplacian pyramid decomposition method, Wavelet transform fusion methodand Contour wave variation method is evaluated objectively.