Infrared And Visible Image Fusion Methods And Evaluation Indexes Research

Infrared&visible image fusion and image fusion quality evaluation are important problems in image processing.In the field of infrared and visible image fusion,it's important to design fusion rules to generate single fusion image by combining the advantages of human vision,with the fusion image obtains some useful information from source images.It's certainly great significance for the development of image processing.The image fusion evaluation index can be used to measure the improvement of fusion image in visual information,spectral information and resolution.Focusing on the infrared&visible image fusion and fusion image quality evaluation,the following work has been completed:(1)The recent research situation of infrared and visible image fusion together with the image fusion evaluation methods are summarized.The infrared and visible image fusion algorithms in the time and frequency domain,the various fusion rules and the development of the algorithms theory are introduced.(2)The multi-sensor image fusion quality evaluation methods are described comprehensively,and the advantages and limitations of different fusion level images evaluation indicators are discussed.The comprehensive table of image fusion evaluation index has been proposed with comprehensive analysis of image fusion evaluation index.According to the feature object and the evaluation purpose analysis,the classification table of image quality evaluation is put forward.(3)The pseudo-color fusion algorithm based on NSCT and adaptive color transfer parameters is presented.By using NSCT and Canny edge detection algorithm to get gray fusion image.Gray fusion image with the source image signal difference are inserted color three-channel respectively to generate the YCbCr source color image.The color transfer model can be used to match the source color image and the target image color statistically.At the same time,the color transfer parameters can be adjusted adaptively because of the color transfer parameter model.Experimentalresults show that the algorithm can provide high contrast and natural color in the false color fusion image.The algorithm can also inhibit the color permeating into the target effectively.The reference image do not need to be strict in this algorithm.(4)The image evaluation algorithm based on the correlation coefficient of multi-scale filter and local variances is presented.Using the Non-Down-sampling filter,the scale analysis is obtained by decomposing the source and fusion image,so the evaluation index can obtain the detailed information in a comprehensive and accurate way.The direction of scale is decomposed by using non-sampling direction filter group,so the evaluation index can extract directional detail information and information in different scales.The final evaluation is obtained by calculating energy information and local variances in each layer combing with the evaluation coefficients.The experimental results show that the performance of the evaluation algorithm is high.