Study On Anterior Eye Alignment Algorithm For Optical Coherence Tomography

On the basis of low coherence interferometry,Optical coherence tomography（OCT）is a new imaging technology with the advantages of no invasion of tissue,three-dimensional imaging in vivo,high-resolution real-time imaging and so on,which has a wide range of applications in the biomedical field.Since ocular tissues have optical permeability and fit with the characteristics of imaging principle of OCT,the field of ophthalmic medicine is an ideal field for the application of OCT technology.When imaging the fundus by OCT,its sample light needs to penetrate through the human eye pupil,go through the lens,and reach the fundus to obtain better quality fundus imaging.But most of the current OCT equipments are operated manually.While aligning the pupils under visible light irradiation generally,the pupils will shrink smaller subjected to visible light irradiation,which results in lower examination efficiency and is not beneficial to imaging by OCT.Therefore,the anterior eye alignment algorithm for OCT is studied in this thesis,which makes the optical axis of OCT sample arm automatically align to the pupil center of eye,improves the automation of OCT equipment,saves time and labor costs,and enhances the imaging efficiency.The main research contents of the thesis are as follows:1.Based on current mainstream eye detection technologies,the integral projection function（IPF）method is applied to human eye detection.In the central detection of the pupil,near-infrared light is used for the light source and an industrial area array CCD with a small field of view is used for collection of images of the eye.Therefor the gray level characteristics and shape of the pupil in the image are more prominent and only contain the human eye and its surrounding skin.Based on these characteristics,firstly,the image is preprocessed to further eliminate the noise effect and highlight the pupil characteristics.Secondly,the image of Region of Interest（ROI）of the pupil is extracted.The method is as follows:the position of the bright spot in the pupil is obtained by combining the Hough transform method detection and IPF,and then the ROI image of the pupil is segmented according to the relationship between the center of the bright spot and the contour points of the connected domain in the binarized image.Finally,the pupil is precisely positioned on the basis of ROI images.The method is as follows:pupil area was segmented from the ROI image of the pupil using the OTSU method,and then locating the pupil center in original image combined with the geometry feature of pupil.2.According to three-dimensional measurement idea of binocular stereo vision,an anterior eye alignment algorithm is designed based on triangle geometry principle,in order that three-dimensional information of the pupil center is obtained from the coordinate information of a pair of two-dimensional images in the image point pupil center.Firstly,the algorithm model and the significance of each parameter are introduced,giving the criteria for alignment as well as the basic process.Then,based on the triangle geometry principle,three components（x₀、y₀、z₀）of pupil center in three-dimensional space are calculated.The x₀ and y₀components are classified and discussed,meanwhile the criteria for determining the pupil center on the X₀ or Y₀ coordinate axis are analyzed.It is proved that the pupil center on the coordinate axis is a special case in the quadrant of X₀-Y₀plane.Finally,a simple anterior eye alignment system is set up for feasibility and anterior eye alignment experiments,and the experimental results achieve the expected effects,which indicate that the anterior eye alignment algorithm is able to be used in OCT for anterior eye alignment.