Self-adaptation magnetic resonance parallel imaging method and device

The invention discloses a self-adaptation magnetic resonance parallel imaging method. The method comprises the steps that the sensitivity initial values of channels are pre-estimated and a dictionary is initialized according to k space data collected in a parallel mode; a target function corresponding to a target image is established according to the sensitivity values and the dictionary, wherein the dictionary comprises a sparse coefficient; the target function is substituted into an augmentation Lagrange frame, interaction updating is carried out on Lagrange multiplier vectors and the target function, and in the interaction updating process, iteration updating is carried out on the sensitivity, and the dictionary and the sparse coefficient are updated; learning is carried out through the updated dictionary, the structure information of the target image is extracted in a self-adaptation mode, and the reestablished target image is obtained according to the structure information and the target function. By means of the method, the structure information of the target image to be reestablished can be caught in a self-adaptation mode, and the sensitivity of each channel can be updated. The invention further provides a self-adaptation magnetic resonance parallel imaging device.