Aircraft Recognition from Satellite Images

In this work, rapid and high-accurate aircraft detection method in satellite images is developed. To this end, AdaBoost learning algorithm is used and Viola-Jones' near real time and high-accurate face detector utilizing haar-like features is taken as reference. In existing studies, for a sub window to be positive, it must pass through all the strong classifiers as positive. However, in this work the strong classifier output values are summed up and compared to a threshold value as well. Therefore, besides the cascade structure's ability to eliminate negative sub windows rapidly, more elaborate evaluation is made on the class of sub window, giving rise to a high performance classifier.