Study On Fusion And De-redundancy Algorithms Of Visible And Infrared Video

In order to solve the problem of insufficient information captured by single sensor,multi-sensor video fusion has become a very effective solution.The fusion of visible and infrared video has always been an important research object in the field of video fusion,and it has a wide range of applications in civil,military,medical and aerospace fields.Visible video contains rich texture information of the target scene,but because the detection distance of visible light is relatively short,and it can not penetrate the physical media such as fog,rain,dust,etc.,visible video is easily affected by harsh environment.Infrared video can use thermal radiation to convert infrared band information beyond human eyes into visible information,and infrared video can recognize hot objects well and detect long distances,but its video contrast is poor and details are not rich.At the same time,video data presents exponential growth,which brings huge challenges to video storage and transmission.In order to solve this problem,it is of great research value to delete the redundant data by utilizing the redundancy of video data.Therefore,this paper designs a video fusion algorithm based on guided filter and weighted two-dimensional principal component analysis by combining the complementarity of visible video and infrared video.By studying the characteristics of video redundancy and combining the redundancy of video,a video de-redundancy algorithm based on average hash is designed.The detailed research contents and contributions of this paper are as follows:(1)Considering the high demand for video quality,a video fusion algorithm based on guided filter and weighted two-dimensional principal component analysis(GW2DPCA)is designed for the complementarity between visible and infrared video.Firstly,the algorithm decomposes the video frames into the basic layer and the detail layer using the guided filter.Then,the adaptive weighted two-dimensional principal component analysis algorithm is used to fuse the basic layers of visible and infrared frames.The detail layer of the visible frame is retained,and the detail layer of the infrared frame is discarded to prevent the real texture from being destroyed by the infrared texture.Finally,the fusion frame is obtained by combining the fused basic layer and detail layer.The fusion result not only preserves the texture information of the visible video,but also preserves the target object of the infrared video.Compared with the existing methods,the fusion frame obtained by this algorithm has more mutual information and higher structural similarity with the original frame,as well as higher overall standard deviation and peak signal-to-noise ratio,and the overall fusion effect is better.(2)Considering the limitation of data storage capacity and equipment energy in the system,this paper designs a video de-redundancy algorithm based on average hash(DRAH)for video redundancy in time.Firstly,the average hash of video frame is calculated,and the Hamming distance between adjacent frames is obtained by comparing the number of different bits in the hash value.Then,it is judged whether two adjacent video frames are approximately duplicated.If they are duplicated,If they are duplicated,the duplicate video frames are deleted to reduce the amount of redundant video data,thus reducing the storage space occupied by redundant data in the system,and reducing the energy consumption required for data transmission.Finally,the experimental results show that the designed algorithm has lower video de-redundancy error rate than the existing methods.When the error rate is the lowest,the algorithm obtains a compression ratio of14.9%,which achieves better video de-redundancy effect.