Research On Similar Face Based On IOS

Face recognition is still emerging as a new field, face recognition as a theme after years of development,Its technical implementation is derived from the three party platform and the platform of the package API function call two aspects,What t can be characterized by matching and find the similarity is a fewl,while the definition of similarity is not the basic standard and algorithm. Most platforms currently used to support the frontal face recognition, and the recognition of the eyes, nose, and mouth position, some of the platform can find a smile face, estimates of age, but the similarity did not support. In order to enable recognition algorithm,it will be a good practical significance by suitable platform.As IOS platform is widely used in this platform to, recognition of similarity will be used in the face to unlock the phone, face payment and other new areas of use,In this paper, we first use the iOS platform hardware as support, the iOS comes with the face recognition API to achieve, and we use OpenCV to achieve the face recognition effect comparison, After finding out the user’s face, we introduce the opencv framework into the iOS development environment Xcode,based on the surf feature matching algorithm Dmatch, we use a method of similarity definition opencv similar recognition software development in iOS platform. At the same time we the add different filters to remove the hair and other effects, using different characteristics of the corresponding algorithm to achieve the best results. Combined with the surf algorithm to we find the most similar to the user’s head portrait.This paper created Dmatch similarity recognition method which is mainly based on the feature points. To characterize the degree of similarity,we try to find the SURF operator, with interfering the filtered among lots number of stablefeature points. Chapter 5 test work is a reliability test, first of all we input a face as a user’s face, which can be found right from the star face library. Then we test the face of scaling and rotation, it still can match the original face from the test,then the user face test and get the matching process map, the similar place of both face are respectively connected and we can find more than 10 characteristics.Then the app outputs star’s face and help users realize their’s the most likely star.In the test process, this paper compares the BFmatch and FLNN filter in their respective effectiveness and adjust their parameters to get the best matching effect, after SURT operator minimum distance parameters in 2000 we reach the best matching effect, and find the feature points, because will also need to connecte feature points. KNN match is a step after a matching connection method,.It can remove the redundant feature points out, so that they can remove interference of hair and face, only the stable points are connected. The connection point of the log with recursive algorithm, can we find the most similar star.