Image Mosaic Research Based On Fast Convergence Ant Algoritlim

Image mosaic technology solves the contradiction between highresolution and wide field of vision effectively, so it has become atechnological frontier in the field of digital image processing. On thebasis of further researching and studying the image registrationtechnology and image fusion technology, the improved algorithms areproposed in view of the insufficiency in the existing algorithms of imageregistration and fusion. Main work includes the following three aspects:(1)Through the further study of the SIFT algorithm, in view of SIFTfeature matching algorithm has more false matching points and theRANSAC algorithm has more iterative times, which lead to the problemsof slow speed and poor effect in image mosaic, the paper improves theSIFT feature matching algorithm and puts forward fast convergence antcolony optimization algorithm--FCACO algorithm to optimize thematching point pairs, through the simulation experiments prove that notonly the improved matching method in this paper eliminates the falsematching points effectively, but also reduces the matching time.(2)In view of the problem that the general linear weight methodproduces the seam-line easily after image fusion, this paper presents asplicing gap elimination method of smooth, gradual weight coefficient.This method firstly adjusts the brightness of the images to make thebrightness consistent, and then improves the general linear weight method,converts the weight of linear mutation function to the form of smooth,gradual cosine function. The experimental results show that the improvedimage pixel gray value transition is smoother, the fusion images have noobvious splicing gaps and the stitching effect is better.(3)Using different kinds of images to make the contrast experimentsbetween the proposed methods in this paper and the existing algorithms,a kind of objective FSFR (Fusion Standard Features Ratio) qualityevaluation index is put forward and then combining with the subjectiveindex to evaluate the algorithms in this paper, results show that not onlythe proposed algorithms in this paper improve the matching precision, but also reduce the time of image mosaic, at last it gets high resolution ratioimages of meeting human’s visual perception, subjective and objectiveevaluation result is consistent.