Research Of Image Registration Technology Based On The Multi-sensor Image Fusion System

In order to improve definition and registration accuracy of the image in multi-sensor image fusion system,an image registration technology based on the multi-sensor image fusion system was researched.Through building a multi-sensor image registration objective evaluation criterion which conforms to the human eye perception,an image registration algorithm which has good stability and low rate of mismatch was proposed.This algorithm was also used in the image fusion system which contained MIR detector,SIR detector and Laser rangefinder.Using this method,fusion image definition can be improved.Due to the lack of the image registration objective evaluation method which conforms to subjective feeling,we researched on the objective evaluation index of multisensor image registration based on boundary definition,target saliency and color consistency.Through the introduction of weber's law,we built the human eye perception model and calculated the human eye perception map of the fusion image.Combined with the local band contrast model,we proposed the boundary definition index of gray fusion image.At the same time,we used the frequency-tuned salient region detection model to calculate the target saliency index of the color fusion image.We calculated the color difference between the fused image and reference color image,and then proposed a color consistency index of image registration.At last,we proposed a new registration evaluation method for multi-model images which has superior performance.The experimental results showed that the new index has higher consistency with the subjective feeling and this index could effectively assess the multi-model images registration accuracy problem in the night vision field.At present,the registration accuracy of multisensor image registration algorithm is not high,and fusion image resolution is poor.Through considering the fusion and registration process together,a joint image registration and fusion method for multispectral infrared images is proposed to obtain the optimal registration parameter for the following fusion process.First,Registration parameters and affine transformation matrixare are defined to regist the SIR image and MIR image.Then Laplace pyramid fusion algorithm is used to fuse the registration image and the reference image.Then definition index of the region of interest in the fused image is used to estimate the fusion image.At last,simulated annealing method is used to find the global optimal solution of evaluation function.The experimental results prove the effectiveness and stability of the method.In order to improve the operation rate of registration algorithm,the front-end optical correction of multisensor image needs to be corrected.Due to the lack of complete optical test method and device,the multi-sensor image fusion optical test equipment was designed to slove follow problems.1.The size of the field between the detectors is not unified.2.The optical axis angle between the detectors is too large.By measuring different size of the field between the detectors,we can correct scaling relation of the multi-sensor image.By measuring the optical axis angle between the detectors,we can correct translation and rotation relationship of the multi-sensor image.By measuring the resolution of multi-sensor image and fused image,we can preliminary estimate the registration accuracy and realize the coarse registration of fusion system.In order to increase the system stability and improve registration precision and quality of fused image,we combined the above three studies and then designed the image fusion system which contained MIR detector,SIR detector and Laser rangefinder.The functions of image registration,image fusion,image matching,target distance detection,target segmentation and tracking are contained by this system.For the convenience of the user operation,the PC control software was designed by us.The experimental results show that the system has highly registration accuracy and good fusion quality.It also has the function of 360°image matching.The flat-fell seam is smooth and without distortion.The system can accurately measure the target distance within 5 km.It also can identify and track the target(3mx3m)within 5km.Development of the system can solve the poor registration accuracy in multi-sensor image fusion system,and it also has important theoretical and engineering significance.