Image feature vector extraction method based on self-organizing mapping and deep learning

The invention discloses an image feature vector extraction method based on self-organizing mapping and deep learning. The method comprises the following steps: dividing a to-be-processed image data set into a training sample set and a test sample set; building a network model of an image feature vector extraction method based on self-organizing mapping and deep learning; constructing a cosine cross entropy classification loss function; constructing a self-organizing mapping objective function suitable for batch learning; combining a cosine cross entropy classification loss function and a self-organizing mapping target function suitable for batch learning to construct a total target function; training the network model constructed in the step 2) by using a mini-batch-based stochastic gradient descent method until the model converges; according to the network model learned in the step 6), the images in the test set are input into the network, and the output of the feature layer is the feature vector of the imag