Novel aircraft type recognition with learning capabilities in satellite images

The collection of satellite images is not constrained by time and can be captured day and night. It is unlike the images captured by aircrafts which are heavily constrained by weather conditions and environmental factors to secure useful images. Recently, satellite images have been widely applied in many fields, such as resource mining, pollution monitoring, etc. In this paper, we plan to apply it to the military to analyze different types of aircrafts for security purpose. In our system, image processing techniques are first employed to perform the image preprocessing tasks, such as image quality enhancement, noise removal, rotation, scaling, and translation adjustment. Then, distinguishable features are extracted from aircrafts for recognition. Last, a multi-level recognition scheme is adopted for recognizing the types of aircrafts by incorporating suitable weight into each recognition level. Experimental results reveal the feasibility and validity of the proposed approach in recognizing aircrafts in satellite images.