Research On Detection Of Corneal Arcus And K-F Ring

Corneal Arcus and K-F(Kayser-Fleisher ring)ring are produced when an abnormality of human lipid and copper metabolism.Detection of the Corneal Arcus and K-F ring has an important significance in screening and diagnosis of people's health condition.The related research shows that the observation of corneal arcus and K-F ring can be used as an assistant method for the timely diagnosis of some diseases.Miscarriage of justice is usually exist when the screening of these two kinds of circular sympthoms is mainly based on observation subjective.The corneal arcus and K-F ring is difficult to disappear even through the treatment.In adddition,the size and the width of thoes two kinds of ring will changed during the metabolism related abnormal.It is impossible to quantify the severity of corneal arcus and K-F ring by the method of subjective observation.The main purpose of this paper is to build a detection model of this two kinds of rings by means of image processing and analysis,which is based on the computer aided detection of the corneal arcus and the K-F ring.In this paper,the main problem is the coarse classification of the target,that is to determine whether there are two kinds of circular features in the image or not.After that,to complete the quantitative detection of the corneal arcus and K-F ring.The specific work and contributions of this paper are as follows:(1)Aiming at the coarse classification of the corneal arcus,we locate the effective area by detection of the eyelashes and eyelids at first.In the next place,the H component and the S component are projected in the unit vector.On the basis of the similarity measure,the Union-Find algorithm is used to complete the clustering of the corneal arcus.Finally,the segmentation of the corneal arcus is complete by the color metric.The experimental results show that our method can reach 95.4% segmentation accuracy when the image is acquisited in eyes opend naturelly and 4.5% miss rateat in the same time.(2)In order to reduce the miss rate of the segmentation by Union-Find algorithm,a corneal arcus segmentation method based on multi-scale color replacement is proposed.Through the analysis of the image in the gallery,the number of targets is stable and the target area is obviously larger than that of the noise.The color replacement strategy is designed by combining the information of neighboring pixels.Firstly,the image is quantized.Secondly,theimage is processed by our replacement strategy in the different scale.The experimental results show that the segmentation accuracy reaches 97.0% respectively.The proposed method is not sensitive to noise and not necessary to locate the iris.(3)To solve the problem of corneal arcus quantitative analysis,we select the candidate area of corneal arcus on the basis of iris edge location.In candidate area,we design the edge enhancement algorithm based on optimal gradient response.The randomness of the inner and outer edges of the corneal arcus in different color gradients is transformed into a uniform distribution,the edge extraction of the corneal arcus ring is completed by combining the boundary tracking and illumination detection model.This method can not only extract the inner and outer boundaries of the corneal arcus effectively,but also achieve good performance for the extraction of the boundary of the K-F ring.Because all images are collected while the participant‘eyes are opened nature,our method still not performance well if the image has the characteristic of error location,eyelid occlusion or doublespot effect.These issues will be the main content of our future research.(4)The degree of corneal arcus and K-F ring was different in different subjects,this difference mainly reflected in the distribution range of circular area.The stability feature quantization operator for color change is defined to the classification of corneal arcus and K-F ring.The width operator is defined by using the edge information of two kinds of rings.The stability and validity of this two kinds of operators are analyzed by the quantitative analysis of the images obtianed at the same person.(5)Aiming at the coarse classification of the K-F ring,locating the ROI area after iris location firstly.Then,we analyze the gray level distribution of different channels in different color spaces and the sensitivity ofillumination in different color space.The HSI is used as the color space for the coarse classification of the K-F ring.In the HSI space,the H components is used for color selection and the S component is used to analyzed the different degrees of K-F ring.Calculating the quadratic transformation of interval H component and S component as discrimination index.The coarse classification of the K-F ring is achieved by discrimination index.