| Leucorrhea is the vaginal discharge of women,and microscopic examination of leucorrhea is one of the current routine clinical examination items.All kinds of biological cells in leucorrhea reflect the pathological changes of female vagina.At present,most of the examination of leucorrhea hospitals are using artificial visual microscope combined with chemical reagent,test,high requirements for inspection the doctor’s subjective judgment of the doctor’s manual,there may be errors.These years,with the digital image processing technology,many areas began to use digital image processing,and the medical field is no exception,digital image processing for medical diagnosis has brought substantial change,so as to promote the reform of medical diagnosis.With the continuous development of intelligent recognition technology,medical diagnosis has gradually begun to identify and analyze the leucorrhea type composition,as far as possible to reduce the frequency of use of medical microscope,which can not only reduce eye fatigue strength of medical inspection personnel,but also to speed up the process of the doctor’s diagnosis.According to the information of the medical circles,the intelligent identification technique of the shape components in the leucorrhea micrograph has not been reported at home and abroad.First of all,the preliminary analysis of the leucorrhea micrograph is carried out,that is to segment the microscopic image and analyze it by morphological element operation and region mark.But the leucorrhea environment is more complex,so in this study,using threshold segmentation,edge segmentation to preprocess the image,so as to carry out assessment of the algorithm according to the processing result from leucorrhea,comprehensive situation,edge segmentation method is better in the microscopic image processing of leucorrhea,so in this thesis.The study mainly adopts the edge segmentation method as the main algorithm of leucorrhea cell extraction,to extract each target area for the next step processing.Secondly,the study of artificial neural network algorithm,identify the main method of white cells is a fuzzy recognition algorithm based on improved neural network,in this study,the fuzzy recognition algorithm using BP(Back Propagation)neural network,and use neural network to fit the membership function.The bacillus and coccus of epithelial cells,with the morphological features and texture features to identify,extract the morphological characteristics,basic geometric gray value,variance,roughness and correlation,combined with epithelial cell image,bacillus and coccus image carries on the introduction of these features.Finally,introduces the whole process of algorithm design,the design of the program,then the program optimization,using CUDA(Compute Unified Device Architecture)to accelerate the image algorithm,finally in order to improve the speed of algorithm,this paper used in the study GPU(Graphics Processing Unit)multi thread to control the operation process.The experimental results show that this algorithm is accurate detection rate of composition recognition type leucorrhea microscopic image reached 94%,the detection speed is 0.26 seconds,which met the requirements of the hospital’s actual testing and has certain application value. |
