Study On Collection?Denoising And Seamless Mosaic Of Pathological Section Images

Limited by traditional collection equipment,the efficiency of cancer pathology analysis can not meet the growing demand for medical treatment.With the further study of computer vision and the rapid development of digital technology,it is of great practical significance to use digital image processing technology to further optimize the processing of collected pathological section images.At the same time,the research on pathological section image processing is still in an immature state at home and abroad.Based on the contradiction of cancer diagnosis needs and the current research status of pathological section image processing,this paper studies the relevant processing of pathological section image.This paper adopts the scientific collection methods to collect the pathological section images,which will be used as experimental materials to carry out the research on image denoising and image mosaic then.The specific contents of the study are as follows:In the denoising stage,this paper introduces the sources and types of noise in images,and the denoising algorithms which used commonly.The denoising principles of Gaussian filter and wavelet threshold method is analyzed in detail,and the shortcomings of the two traditional algorithms in denoising pathological section images are analyzed.Therefore,an improved algorithm based on the traditional wavelet denoising principle is proposed in this paper.According to the construction method of threshold function summarized in a lot of literatures,this paper proposes a new threshold function.It proves that the improved function makes up for the defect of the traditional wavelet soft and hard threshold function by means of mathematical derivation.At the same time,Matlab simulation experiments are carried out to prove the superiority of the improved threshold function denoising performance.At the stage of feature matching,this thesis focuses on the SIFT feature matching algorithm based on scale space.Firstly,the basic principle and key steps of SIFT feature matching are analyzed in detail,and then the improved SIFT matching algorithm is proposed by studying the relationship between image entropy and feature points.The improved matching algorithm eliminates unstable SIFT feature points and the matching pairs with large entropy difference,but retains the matching pairs with strong stability and high precision,which can achieve the effect of accelerated matching.Finally,the simulation results show that the improved matching algorithm not only ensures the robustness of matching,but also improves the speed of feature matching.After the feature matching is completed,the appropriate spatial transformation model which according to the features of the collected pathological section images is selected to align the images to be spliced.Finally,the average fusion method and the gradual-in and gradual-out fusion method are combined to eliminate the trace of the mosaic image and complete the seamless mosaic of the pathological section image.