Research Of Multi-sensor Image Fusion Algorithms Based On Multi-resolution Analysis

Image fusion is an integrated information processing technology that synthesizes several image of the same scene or target to improve the accuracy,integrity and credibility of the image description.It significantly improves the probability of target recognition and track under complicated and disturbing conditions,so as to lay a foundation for the accurate positioning and accurate strike of the target.This paper mainly studies the multi-sensor image fusion technology based on multi-resolution analysis.The fusion result obtained by this technique is more in line with the human visual perception and is more conducive to the further analysis of images.The main research contents are as follows:1.For multi-resolution analysis,the performance and features of several commonly used multi-resolution image fusion algorithms based on Laplace pyramid decomposition,wavelet transform and contourlet transform are systematically analyzed,and the research ideas are rationalized for the subsequent research work.2.The principle and algorithm of nonsubsampled contourlet transform(NSCT)are studied in depth.According to the imaging mechanism of SAR image and visible image,a SAR image and visible image fusion algorithm based on nonsubsampled contourlet transform is proposed.For low frequency sub-images,neighborhood entropy is used as the measure parameter for neighborhood fusion.The high frequency sub-images are fused by the neighborhood algorithm based on the average gradient selected and by using the neighborhood correlation coefficient as the threshold.Experimental results show that the algorithm can significantly improve the visual effect of the fused image.3.In view of the characteristics of infrared and visible image fusion,an infrared and visible image fusion algorithm based on nonsubsampled contourlet transform and hybrid particle swarm optimization algorithm is proposed.For the low frequency sub-images,an improved weighted average method based on regional averages is adopted.For the high frequency sub-images,the threshold value is selected by the hybrid particle swarm optimization and the neighborhood algorithm based on the average gradient selection is adopted.Experimental results show that the algorithm can effectively improve the fusion of fusion images.4.Using the statistical characteristics of the images as the criterion for evaluating the performance of fusion,the experimental results are quantitatively analyzed.The results show that the fusion algorithm can capture the details such as edge and texture more effectively,and obtain better fusion effect.