Deep learning sample preparation method and identification method based on satellite orthoimage full attitudes

The invention relates to a deep learning sample preparation method and a target identification method based on satellite orthoimage full attitudes. The deep learning sample preparation method comprises the following steps of: (1) generating target region images ( as well samples to be involved in deep learning training) of an aircraft and a target under different distances, different azimuth angles and different elevation angles by using satellite orthoimage data of the target and using a ray tracing method; (2) performing gray scale inversion processing on the target region image to obtain asample; (3) performing Gaussian blur processing on the target region image to obtain a sample; (4) performing logarithm transformation on the target region image to obtain a sample; (5) adjusting thebrightness of the target region image to obtain a sample; (6) carrying out histogram equalization on the target region image to obtain a sample; and (7) forming the samples in the steps (1)-(6) into afinal sample set to be inv