Research On Image Processing Technology Of Pathological Section Digital Scanning Equipment

Pathological section is an important criterion for disease diagnosis.But observing pathological sections by the traditional microscope is inefficiency for pathological diagnosis and pathology education.The pathological section digital scanning device collects all the magnifying partial images of Pathological section with the cooperation of camera and motion platform,and then stitches these partial images into a whole slide image by using image processing algorithm on the computer.Scanning a Pathological section into a digital slide need to get and stich thousands of microscopic images.So the performance and speed of image processing algorithm have great influence on the quality of digital slices and the scanning efficiency of the scanning device.In this paper,the researches on microscopic images mosaicking and multi-focus microscopic image fusion of the pathological section digital scanning equipment are carried out.The main contents are as follows:(1)A method for target extraction of pathological sections is proposed to locate the key area for scanning.By reducing the slice scanning area,the scanning efficiency of pathological sections scanning device is improved.(2)The auto-focusing method to get a clear microscopic image is achieved by using the Laplacian gradient function as the sharpness evaluation function.And a pre-focus technique is adopted to speed up the scanning by reducing the times of focusing during the acquisition of microscopic images.The focus position of each microscopic image is estimated by the value of the nearby Latin sampling points directly,which is the main idea of pre-focus technique.(3)The registration of adjacent microscopic images is realized by grayscale template matching.For the problem of slow registration of grayscale template matching,an improved grayscale simplified template fast registration method for the sequential microscopic images is proposed.According to the characteristics of strong correlation of adjacent microscopic image acquisition locations,an inertial search strategy is proposed.And local upsampling phase correlation method is used on the subgraphs of microscopic images to obtain sub-pixel level registration parameters.(4)For the problem that the gradual fade-out fusion algorithm can not mitigate the overall brightness difference of the fused image,a reference weight correction method based on HSV color space is proposed.(5)For some thicker pathological sliced tissue regions,the clear full-focus microscopic image of these regions is acquired by using Shearlet transform to get the multi-focus images fused,which are grabbed from different focal plane of these regions.And according to the characteristics of microscopic images,improved fusion rule is proposed.(6)The working principle and structure of the pathological section scanning device is analyzed in detail.On this basis,the prototype pathological section scanning system and software which are used to verify the effectiveness of the above algorithms,is designed and established.