A cervical exfoliated cell nucleus image recognition method

The invention relates to a cervical exfoliated cell nucleus image recognition method, comprising the following steps: 1, inputting training and testing sample images of the cervical exfoliated cell nucleus; 2, preprocessing the image of the cervical exfoliated cell nucleus; 3, extracting features of...

Full description

Saved in:
Bibliographic Details
Main Authors XIA QIKAI, LI QINGHUA, WANG ZHE
Format Patent
LanguageChinese
English
Published 18.12.2018
Subjects
CALCULATING
COMPUTING
COUNTING
HANDLING RECORD CARRIERS
PHYSICS
PRESENTATION OF DATA
RECOGNITION OF DATA
RECORD CARRIERS
Online AccessGet full text

Cover

More Information
Summary:The invention relates to a cervical exfoliated cell nucleus image recognition method, comprising the following steps: 1, inputting training and testing sample images of the cervical exfoliated cell nucleus; 2, preprocessing the image of the cervical exfoliated cell nucleus; 3, extracting features of the cervical exfoliate cell nucleus, normalizing and labeling the sample image; 4, constructing SVMclassifier with mixed kernel function by weighted summation; 5, using large mesh and QPSO algorithm to search for the best region; 6, using middle mesh and QPSO algorithm to optimize the optimal region and reduce the optimal region; 7. using small mesh and QPSO algorithm to find the optimal mixed kernel function parameters and establish the optimal SVM classifier; 8. Identifying and classifying the cervical exfoliated nuclei of the test set by the optimized new SVM classifier. The method recognizes the cell types in different nuclear images and achieves the effect of fast nuclear recognition. 本发明涉及种宫颈脱落细胞核图像识别方法，步骤：1
Bibliography:Application Number: CN201810555507