Research On Classification And Recognition Of Milk Somatic Cells Based On Improved Features

Mastitis is a serious disease that harms dairy farming.Research shows that the type and quantity of milk somatic cells is an important indicator for the evaluation of milk quality and the diagnosis of mastitis.There are four types of milk somatic cells:neutrophils,lymphocytes,macrophages,and epithelial cells.When the cows suffering from mastitis,the number of somatic cells has been greatly increased.Through computer technology,it can realize classification and counting of milk somatic cells.According to the number of four types of milk somatic cells,it can detect the extent of mastitis infection,and is covenient to take timely treatment programs to prevent mastitis spreading.Therefore,in order to solve some problems in the examination of milk milk cells,to improve the efficiency and accuracy of mastitis diagnosis,the study has been conducted on the texture and morphological characteristics of milk somatic cells.The main contents are as follows:(1)Pretreatment operation of cells.Owing to some casein,debris,lactoal-bumin in somatic cell images,and different stains produce noise,some pretreatment operations are needed.First cells images were pretreated by gray processing,binarization,wave filtering,mathematical morphology.And then,the area-of-interest was obtain by K-means clustering segmentation algorithm,to provide a basis for subsequent experiments.(2)The milk somatic cell classification and recognition algorithm based on improved morphological characteristics is proposed.First,the texture features of the four types of cells were extracted.Then,morphological features were extracted from the segmented cells and nuclei.According to the shape characteristics of the four types of cells,the expressions of roundness and rectangularity have been improved.Experimental results indicate that the improved morphological features have higher recognition rate,the random forest classifier can also be used to get ideal classification results.(3)The milk somatic cell classification and recognition algorithm which fused LBP with GLCM is proposed.LBP is able to descride the local texture details.Moreover GLCM focused on more global texture information.The fusion of LBP and GLCM can get better results in identifying somatic cells.