Multiresolution Robust Image Fusion Algorithm And Its Application In Intelligent Perception Of Unmanned Vehicles

Recently,the technology of sensor is updated quickly.Image information acquired through the technology is becoming more and more,which puts forward higher requirements for the technology of image processing.Among the numerous source images acquired,it is inevitable that there will be defects in some images,such as the damaged area of image,poor brightness and so on.This paper studies the restoration algorithm for defective multi-resolution images fusion primarily and its application in intelligent perception of unmanned vehicles.First of all,this paper introduces the application fields of image fusion briefly,summarizes the common image fusion preprocessing methods,the level of image fusion,the classification of fusion algorithms and the common objective evaluation metrics.Then this paper proposes an image restoration fusion algorithm based on Kalman filter and a fusion algorithm based on wavelet transform and multi-scale decomposition of guided filter.Image restoration fusion algorithm based on Kalman filter is mainly used to repair the visible image with damaged area.This algorithm firstly determines the location of the damaged region by edge detection in gradient domain,and then repairs the defect area by block fusion algorithm based on Kalman filter with changing the innovation value of the damaged area.Finally,fusion images will eliminate the effects of damaged areas greatly.The algorithm is simulated by the source image captured by the unmanned vehicle,which proved its effectiveness.The wavelet transform algorithm based on multi-scale decomposition of guided filter achieves enhancement and repairing of visible images in night vision by introducing corresponding infrared images.The algorithm firstly uses wavelet decomposition to make the source images have the same resolution components,then enhances the low frequency components of the visible image through adaptive enhancement algorithm based on guided filter,finally uses guided filter multi-scale decomposition and weighted average fusion as the fusion rule of wavelet transform to achieve the fusion of visible image and infrared image.Simulation experiments show that the above algorithm can obtain good fusion results.