The invention discloses a
lung cancer image fine classification method based on fusion of LBP and
wavelet moment features. The
lung cancer image fine classification method is characterized in that S1, nidus positioning of an input image can be carried out; S2, a plurality of templates of the nidus part can be generated randomly; S3, the scaling of the input image in different dimensions can be carried out, and the extraction of the
textural feature MB-LBP and shape feature
wavelet moment of the image block and the template block can be respectively carried out, and then the two features can be integrated by adjusting the weighting parameters by the experiment; S4, the feature response image can be acquired by matching the different positions of the image; S5, the response image can be converted into the
feature vector by adopting the improved mean space
pyramid model; S6, the fine classification can be realized by adopting the vector
machine. The
algorithm provided by the invention is the attempt of the fine classification idea in the
medical science field, and the generating of the redundant templates can be reduced. The
lung cancer image information can be represented by the well fusion of the LBP textural features and the
wavelet moment features. By adopting the
pyramid model extraction model, the strong features can be maintained, and the identification precision can be improved.