Research On Straylight Elimination Algorithm For Fundus Images

Fundus photos are an important basis for the diagnosis of ophthalmic diseases.The new Line-Scanning Ophthalmoscope is susceptible to environmental influences when taking pictures.There may be straylight or even obvious bright spots in the photos taken,which causes trouble when doctors diagnose patients.At present,the main methods for eliminating the effect of straylight are to reduce the possibility of the straylight to be taken into the lens when designing the path of light,and exclude the interference of unwanted light sources when taking fundus photos.This paper proposes an algorithm for processing the original images(video)obtained by the camera to generate fundus images without straylight.The algorithm mainly includes two parts: the first part is to process a single picture,locate straylight and determine the area to be repaired.Wavelet coefficients of the image are calculated,and the information of the area to be repaired is restored based on the corresponding adjacent points on the wavelet coefficients,and then the inverse wavelet transform is used to obtain the repaired image.Finally,we use Gaussian blur to improve the edge of the area.The second part is the fusion of consecutive frames.We determine the translation between frames by calculating the cross-correlation in the Fourier frequency domain.After registration,image fusion is performed in the wavelet domain to obtain the final result.Masks of damaged area are used to reduce straylight during fusion.The algorithm not only retains the original information of the image,but also eliminates noise and uses the image information to restore the area affected by straylight.The final result obtained has better visual effects than the original image.