Research On Segmentation And Recognition Algorithms Of Cervical Cell Images

In recent years,with the increasing incidence of cervical cancer,the method of manually discriminating cervical cell images not only consumes time,but also has a large workload.With the development of artificial intelligence,cell image recognition methods based on machine vision have attracted the attention of many scholars.The key to cell image processing is the extraction of cell edges,and there are complex phenomena of overlapping and fine particle impurities in the cell image.The machine identification method cannot obtain the precise edge of the cell,or obtains false edges.Therefore,it is necessary to choose a suitable contour extraction algorithm.In addition,the traditional image recognition method needs to extract a large number of image features,the redundant features not only increase the calculation amount but also easily introduce interference.In response to these problems,this paper proposes an improved AGVF Snake(Adaptive Gradient Vector Flow)model for the extraction of cell morphological features.Firstly,the improved Canny is used to initially locate the cell image,and then adaptive initial contour model and gradient vector model are used to obtain accurate cell edges to better distinguish normal and abnormal cells from morphological features.In order to extract accurate and effective features for cell identification,this paper proposes a machine learning method based on feature selection algorithm for cervical cell classification research.Firstly,this paper introduces Classification and Regression Trees(CART)for cell feature selection,which reduces the dimension of input feature attributes.Secondly,this paper uses Particle Swarm Optimization(PSO)to optimize the hyperparameters of the Support Vector Machine(SVM),and establishes the PSO-SVM model to classify cervical cancer cells.Finally,the Herlev dataset is introduced to establish a classification model of cervical cancer cells,and the accuracy of cervical cells is observed.Six classification and diagnosis methods of cervical cancer cells are introduced to simulate and compare.The simulation results show that this algorithm can extract accurate and effective features and has high recognition accuracy.In addition,based on the Inception V3 network structure,this paper proposes a method combining convolutional neural network and artificial feature extraction to classify cervical cells,achieving a classification effect of more than 98%.The proposed method has less algorithmic complexity,simple structure,and high accuracy,which can be further extended to the classification and application of cervical cancer cells,thus providing an effective method framework for the diagnosis of cervical cancer diseases.