Automatic Detection Method Of Liquid-based Producer Of Quality Cervical Specimens

Quality control system is important to any system, small or big. The liquid based specimens involves sample collecting, centrifuging, dyeing, fixing, mounting drying. Any minor anomaly will affect the later image automatic detection. So this paper studies the method of image-content-based anomaly detection.With the unqualified specimen specified in 2001 TBS system, based on the visual attributes of the images, rough sets was introduced and the detection model is established and the corresponding detection methods were represented in the paper. In the end all the algorithms of the detection system is rewritten into the OpenCV. The following are the details:(1) comparison of the unqualified specimens specified in the 2001 TBS system with the normal specimens lead to the finding of the perspective characteristics. image processing techniques are used in automatic extraction of image features. Set and rough set theories were introduced and the scope of parameter of the specimen was analyzed before detection model was established.(2) image content description, the detection methods and the detection model were established in the paper. Rules are established to discriminate the coverslips foreign body cover and mounting defects, serious pollution of blood, cell aggregation and inflammatory exudate coverage serious sample image and the specimens of normal image and verified by the experiments.(3) detection model represented in the paper was used to discriminate the unqualified specimens. The rules are established and verified in the experiments.(4) The MATLAB based detection system was rewritten in OpenCV to verify the consistency of the results and to improve the processing speed of the system. The whole system was tested and verified.Now the automatic detection method of liquid based cervical specimens is used in Cuality Control Systems of Cervical Cancer Detecting System in Nanchang CytoDiagnosis S&T Company Limited.