Image Fusion Algorithm Research Based On Local Characteristics

With the rapid development of information, People pay more and more attention todigital image processing technology, Image fusion technology has become one of themost important means of image processing technology, has been widely used in military,target recognition, security, medical diagnosis, etc.At present，though image fusion algorithm is various，but there exist many problems:with relatively simple algorithm, Processing speed is fast，but it’s hard to get goodfusion effect；with complex algorithm, it has Large amount of calculation，andprocessing speed is slow, it’s difficult to meet the requirements of the actual system inreal time；there are more and more images，different images for different algorithms,how to select the most appropriate algorithms is also a problem at present; For imagefusion quality evaluation there isn’t a set of complete and reliable evaluation method offusion. Based on the above problems, this paper conducted the following research work:First, this paper delves into image fusion algorithm，according to the principle oflocal image feature extraction, a image fusion algorithm based on local features hasbeen proposed in this paper, the algorithm based on the structural feature of the imageitself, using the seed point of region growing to locate the image clear area, with spatialfrequency as a threshold for locating clear area with accurate position,finally throughregional substitution to realize image fusion, and write the program code to doexperimental simulation, to prove the feasibility and correctness of the algorithm, thenexpand the way of image fusion algorithm.Second, this paper based on the principles of the local characteristics of image, by comparing and analysing the key points between images, the image local features wascombined with image fusion quality evaluation, a kind of image fusion qualityevaluation method based on local characteristics has been proposed. And write theprogram code to do experimental simulation, to prove the feasibility and the correctnessof the proposed algorithm, so as to expand the way of fused image quality evaluationmethod.