Research On Stereo Matching Algorithm In Laparoscopic Surgery Images

Compared with traditional laparotomy,laparoscopic surgery has many advantages of light trauma,little pain,and little recovery time,etc,which plays an important role in the field of medical surgery.In clinical surgery,Laparoscopic can display depth information,and provide a relatively natural three-dimensional vision for doctors.It makes specific precision faster and more accurate,meanwhile,it reduces the rate of surgical errors significantly,and provides advanced technical support for more complex minimally invasive surgery.Therefore,the study of three-dimensional laparoscopic technique has important theoretical value and practical application prospects.Based on binocular stereo vision principle,the author had done amount of researches on matching algorithm and depth information calculation.The main contents of this thesis are as follows:Firstly,3d laparoscopic surgery system and the principle of depth estimation which based on Binocular Stereo vision are introduced in the paper.Secondly,the advantages and disadvantages for the algorithm of classic Harris feature matching are described in detail.From Harris corner detection operator,the description of feature vector and feature matching algorithm to improve the classic algorithms,and proposes a Harris feature matching algorithm with scale invariant characteristic.Experiments in MATLAB 7.0 software platform,proves the superiority of the algorithm with scale invariant characteristic is effective,and achieve the desired effect.Finally,according to the advantages and shortcomings of area matching and feature matching,a mixed matching algorithm combined Harris feature matching and area based matching was proposed.This hybrid matching algorithm not only has better robust at feature points matching but although can get dense disparity.The improved Harris corner matching algorithm is used in the preliminary matches to speed up computations.Then the disparity gradient constraint area matching algorithm is used to process a large number of other non-feature point.Ultimately,obtain a dense disparity map though extracting depth information by triangulation.On this basis,using binocular vision imaging system,that two cameras' main optical axis are parallel,with non-contact triangulation method to get thedepth of the laparoscopic surgery information.The experiments in MATLAB 7.0 software platform shows that,the hybrid matching algorithm has a good matching accuracy at disparity discontinues edge,and produces a dense disparity map.On resulting depth image,the outline of each object is relatively clear and easy to distinguish,distance relationship on grey value can be reflected well..