Application Research Of Image Classification And Recognition Method Based On Factor Space

As people pay more and more attention to artificial intelligence,image classification and recognition technology,which is one of important branches in artificial intelligence,has evoked more and more attention by researchers.Meanwhile,the causality analysis method proposed by factor space theory in data science has gradually gripped people's attention.Therefore,the study of image classification and recognition based on factor space theory is a novel and meaningful research direction.This paper focuses on the construction and application of image classification and recognition method within factor space theory.The major findings are as follow:Firstly,this paper introduces the principle of image classification and recognition technology and factor space theory,and discusses the possible problems when causality analysis of the factor space is applying in image classification and recognition: 1.If the causality analysis method in factor space is used in incomplete data sets,unidentified problems might occur in some samples.2.The current factor analysis method only discusses the situation of discrete data,since some image classification data sets are continuous,it cannot solve the problem of image classification and recognition by using causality analysis method directly.Therefore,this paper improves the causality analysis method within factor space for the sake of solving the two problems mentioned above.Secondly,for the unidentified samples problem occurs in incomplete data sets,this paper improves the process of classification and recognition of causality analysis in factor space,and proposes an improved causality analysis method based on the principle of proximity.The first step is to set the first recognition on unrecognized samples with the nearest neighbor principle;the second step is to identify the non-unique nearest neighbor in the first recognition with the maximum subordination principle.Thus,all the unrecognized samples are classified effectively.In this paper,the method of factor analysis based on the principle of proximity is used to do the classification test in the Breast Cancer Wisconsin(Original)Data Set.The result shows that all the samples can be identified successfully,and the accuracy rate of the classification is 98.67%,which is increased by 5.34% compared to the previous causality analysis method.Finally,for the applicability problem of factor space theory arises in continuous data sets,this paper proposes a method of image classification and recognition based on factor space,combining data discretization method and causality analysis method based on the principle of proximity.The result shows that the classification accuracy of the method of image classification and recognition based on factor space is 89.29%,which is higher than the four common image classifiers: Navie Bayes,C4.5,SVM and BP Neural Network.The result indicates that the method of image classification and recognition based on factor space proposed in this paper is applied effectively in the continuous data set such as image classification data sets;in addition,the algorithm works well.The theory of factor space provides a feasible new research idea for the field of image classification and recognition.