Research On Infrared And Visible Image Fusion Method And Its Application In Power Equipment Monitoring

With the increasing development of sensor technology,a single imaging sensor is difficult to meet the needs of daily life.Multi-source sensor is becoming the mainstream technology in the world.But at the same time,multi-source imaging sensor technology also brings many problems and great challenges,such as image registration,image fusion and so on.Infrared and visible images are the most widely used images in the field of image processing.Because their imaging principles and imaging characteristics are naturally complementary,their image registration and image fusion tasks have attracted extensive attention.In order to further improve the quality of image fusion,this paper makes an in-depth study on the problems of low registration accuracy and unsuitable fusion strategy due to less feature points or low utilization of image residual information in infrared and visible image registration and fusion,and gives a perfect image registration and fusion algorithm.The main work includes:(1)Aiming at the problems of scale invariance and low registration accuracy in orb algorithm,an improved orb image registration algorithm based on Canny edge detection is proposed.This algorithm first uses Canny operator to detect the edge of the image,then combines surf and fast feature extraction algorithm to extract the feature of the image,and uses NCC algorithm to screen the coarse matching,finally uses PROSAC to get the fine matching result.The experimental results show that the infrared and visible image registration algorithm has achieved outstanding performance in both subjective and objective evaluation,and more accurate registration.(2)Aiming at the problems of texture detail information loss and artifacts caused by inappropriate fusion strategy in traditional image fusion algorithms,a dual channel PCNN fusion framework based on phase consistency is proposed.Firstly,NSST is used for multi-scale decomposition into low-frequency subband and high-frequency subband.For the low-frequency subband,the phase consistency weight map is used for fusion,which makes full use of the contour information in the low-frequency subband.The fusion strategy of dual channel pulse coupled neural network is adopted in the high-frequency subband,and its link strength is improved through phase consistency to realize the effective fusion of high-frequency information.Finally,the fused image is obtained by inverse transformation.Experimental results show that the fusion algorithm has excellent image fusion quality and contains richer edge texture details.(3)Aiming at the inaccurate registration and fusion of power equipment monitoring images caused by sensor position differences,weather interference and other factors,the improved ORB registration algorithm based on Canny edge detection and the dual channel PCNN fusion algorithm based on phase consistency are applied to the power equipment scene.Surf and FAST can effectively extract the feature points in the electrical equipment image for matching,and the dual channel PCNN algorithm based on phase consistency can make full use of the detail information in the high-frequency subband to achieve a good fusion between the two.Experiments results show that the registration and fusion framework can effectively register and fuse power equipment images.Experiments show that the registration and fusion framework can effectively register and fuse power equipment images.