The invention discloses an
optic nerve image enhancement and
lesion recognition method based on multi-
modal fusion, and the method comprises the following steps: S1, collecting the image data of an
optic nerve lesion patient, and carrying out the denoising
processing, spatial position alignment and data normalization; s2, constructing an improved conditional
diffusion probability model, and establishing a
reverse diffusion process for image detail enhancement; s3, gradually enhancing image data details by using a
reverse diffusion process; s4, calculating high-dimensional
conditional probability feature joint distribution of different
modal image space domains and frequency domains, and establishing a cross-
modal feature set; s5, performing image adaptive fusion by adopting a conditional guidance mechanism; s6, optimizing
model parameters to obtain a fused image; and S7, identifying the
feature vector of the
lesion area by using a Transform classification model to realize
lesion identification and classification diagnosis. According to the invention, the
image fusion quality and the lesion diagnosis precision are obviously improved.