The invention relates to a multilayer parallel network SAR image airplane object detection method based on parts, and the method comprises the following steps: constructing a root detection network for detecting a whole airplane and a part detection network for detecting the subcomponents of the airplane; preparing a training data set, marking the whole airplane and the subcomponents of the airplane in the training data set through a surrounding frame, inputting a training image and the corresponding marking information into the root detection network and the part detection network for the training of the networks; carrying out the initial detection of a to-be-detected image through the trained root detection network and the part detection network; carrying out the matching of a root detection result and a part detection result, and obtaining subcomponents matched with the whole airplane; combining prior structure information and detection capturing probability obtained through initialdetection, taking the maximum detection capturing probability as the rule, carrying out the constraint processing of the root detection result and the part detection result, and obtaining a final detection result. The method effectively reduces the detection leakage rate and the false alarm rate, and improves the detection accuracy.