The Research And Application Of Image Matching's Algorithm

Image matching is an important technology in the field of image information processing.An intensive study aiming at kinds of method of image matching has be done in this paper.In this paper there are mainly two kinds of method of image matching being researched:method based on the gray-value of pixels and method based on the features.There are kinds of method of image matching being describing in this paper,and do the experiment for some classical and mature methods. Proposed the improvement ideas and direction of image matching algorithmm through analyzing theirs characteristics. The paper proposed a method which can solve the problem belonged to the method which named template matching.The method import the histogram as the"unit"when do the matching.There are particular it's analyse of performance,merit and defect,the instance it apt to apply, in the paper. A improved SIFT(SC-SIFT),based on separated dimension searching and cricoid feature descriptor,was proposed to deal with the poor efficiency in the Maxima and minima point searching and feature compute.The searching method in new algorithmm works in two dimension first,and then in the third dimension,which in the original algorithmm is working in three dimension once. A new cricoid feature descriptor used in the new algorithmm instead of the original descriptor which is high dimension but low efficency.Experimental results show that the new algorithmm can not only improve the execution efficiency of SIFT,but also raised the matching accuracy, optimize the SIFT algorithmm.Because there often encountered the face of the specific application of retrieval and inquires in the information retrieval field,a matching algorithmm which combining classcal adaboost algorithmm and template matching with histogram to process matching of face detection specificly,is proposed in this paper,and through the experiment,it has proved that in a certain limit conditions, this algorithmm is superior in the speed and accuracy,and thereby it can enhance the reliability and practicability of the relevant system.