Font Size: a A A

Background Analysis Methods And Application Of Cervical Cell Morphological Detection

Posted on:2019-04-19Degree:MasterType:Thesis
Country:ChinaCandidate:C ChenFull Text:PDF
GTID:2334330566458275Subject:Control engineering
Abstract/Summary:PDF Full Text Request
This paper takes background images of cervical cells as the main research object.In this paper,the complicated background in cervical cell background was defined and classified,and the background analysis method in cervical cell morphological detection was studied.In the study of cervical cell background,we collected enough images of cervical cell background and intercepted the cells and blocks in the background.First,to extract the characteristic parameters of the corresponding cells and blocks,the basal layer and subbasal layer cells were detected by fuzzy clustering,and the coverage background of the block was segmented by multi-truncated segmentation.Because of the complexity and diversity of cervical cell background,a set of procedures for detecting the background of basal layer,subbasal layer and block covering were designed according to the sample,which is to identify the cervical cells background quickly and accurately.(1)To analyze the origin and physical characteristics of the mesoscale cells,basal layer cells,subbasal layer cells and blocks,to obtain the visual characteristics of this cells,and to compare and analyze the visual differences among the various types of cells.(2)To detect identify the background image of basal and subbasal cells,the background cells were detected by fuzzy clustering model from the angle of morphology,the background of basal and subbasal cells was analyzed,and the whole background image was identified by using multi-pattern recognition method.For the image analysis of block cover,the type of block is classified first.Because of the inconsistency of block type,the multi-block segmentation model is established.The characteristic parameters of the block are used to segment the block background and the detection algorithm of the block background is obtained.(3)In this paper,the multi-mode recognition method adopted can improve the accuracy of background discrimination and reduce the leakage rate.Compared with the single threshold segmentation method,multi-block segmentation method can be increased and the area of the segmentation can be extended.(4)To verify the validity of the analytical method of the background of basal layer and subbasal layer cell and block.For the background image of basal layer and subbasal layer,we need to check the accuracy and error detection rate of basal layer,subbasal cell and the base layer,subbasal layer cell background.For block coverage background image,we need to comparise the accuracy and false positive rate of block segmentation by multiblock segmentation method,and then to analyze the block coverage background discrimination accuracy and false positive rate.
Keywords/Search Tags:cervical cell background, intelligent recognition, fuzzy clustering, morphological detection
PDF Full Text Request
Related items